Intelligent multimedia databases and information

retrieval advancing applications and

technologies 1st Edition Li Yan download

https://ebookname.com/product/intelligent-multimedia-databases-
and-information-retrieval-advancing-applications-and-
technologies-1st-edition-li-yan/

Get Instant Ebook Downloads – Browse at https://ebookname.com

Instant digital products (PDF, ePub, MOBI) available
Download now and explore formats that suit you...

Learning to Rank for Information Retrieval and Natural

Language Processing 1st Edition Hang Li

https://ebookname.com/product/learning-to-rank-for-information-
retrieval-and-natural-language-processing-1st-edition-hang-li/

Distributed Artificial Intelligence Agent Technology

and Collaborative Applications Advances in Intelligent
Information Technologies 1st Edition Vijayan Sugumaran

https://ebookname.com/product/distributed-artificial-
intelligence-agent-technology-and-collaborative-applications-
advances-in-intelligent-information-technologies-1st-edition-
vijayan-sugumaran/

Intelligent Transport Systems Technologies and

Applications 1st Edition Asier Perallos

https://ebookname.com/product/intelligent-transport-systems-
technologies-and-applications-1st-edition-asier-perallos/

Cop Killer Martin Beck Police Mystery 9 2nd Edition Maj

Sjöwall

https://ebookname.com/product/cop-killer-martin-beck-police-
mystery-9-2nd-edition-maj-sjowall/
Atlas of peripheral nerve surgery 2nd ed. Edition
Hudson

https://ebookname.com/product/atlas-of-peripheral-nerve-
surgery-2nd-ed-edition-hudson/

Emergencies in Trauma 1st Edition Aneel Bhangu

https://ebookname.com/product/emergencies-in-trauma-1st-edition-
aneel-bhangu/

No Regrets 1st Edition Michele Ann Young

https://ebookname.com/product/no-regrets-1st-edition-michele-ann-
young/

Qualitative Research Methods In Business Techniques for

Data Collection and Analysis 2024th Edition De Sordi

https://ebookname.com/product/qualitative-research-methods-in-
business-techniques-for-data-collection-and-analysis-2024th-
edition-de-sordi/

Persons Humanity and the Definition of Death 1st

Edition John P. Lizza

https://ebookname.com/product/persons-humanity-and-the-
definition-of-death-1st-edition-john-p-lizza/
The Catholic Studies Reader 1st Edition James Fisher

https://ebookname.com/product/the-catholic-studies-reader-1st-
edition-james-fisher/
Intelligent Multimedia
Databases and
Information Retrieval:
Advancing Applications and
Technologies
Li Yan
Northeastern University, China

Zongmin Ma
Northeastern University, China
Senior Editorial Director: Kristin Klinger
Director of Book Publications: Julia Mosemann
Editorial Director: Lindsay Johnston
Acquisitions Editor: Erika Carter
Development Editor: Michael Killian
Production Editor: Sean Woznicki
Typesetters: Adrienne Freeland, Jennifer Romanchak, Mackenzie Snader, Milan Vracarich Jr.
Print Coordinator: Jamie Snavely
Cover Design: Nick Newcomer

Published in the United States of America by

Information Science Reference (an imprint of IGI Global)
701 E. Chocolate Avenue
Hershey PA 17033
Tel: 717-533-8845
Fax: 717-533-8661
E-mail: [email protected]
Web site: http://www.igi-global.com

Copyright © 2012 by IGI Global. All rights reserved. No part of this publication may be reproduced, stored or distributed in
any form or by any means, electronic or mechanical, including photocopying, without written permission from the publisher.
Product or company names used in this set are for identification purposes only. Inclusion of the names of the products or
companies does not indicate a claim of ownership by IGI Global of the trademark or registered trademark.

Library of Congress Cataloging-in-Publication Data

Intelligent multimedia databases and information retrieval: advancing
applications and technologies / Li Yan and Zongmin Ma, editors.
p. cm.
Includes bibliographical references and index.
Summary: “This book details the latest information retrieval technologies
and applications, the research surrounding the field, and the methodologies
and design related to multimedia databases”--Provided by publisher.
ISBN 978-1-61350-126-9 (hardcover) -- ISBN 978-1-61350-127-6 (ebook) -- ISBN
978-1-61350-128-3 (print & perpetual access) 1. Multimedia systems. 2.
Databases. 3. Intelligent agents (Computer software). I. Yan, Li, 1964- II.
Ma, Zongmin, 1965-
QA76.575.I254 2012
006.7--dc23
2011027682

British Cataloguing in Publication Data

A Cataloguing in Publication record for this book is available from the British Library.

All work contributed to this book is new, previously-unpublished material. The views expressed in this book are those of the
authors, but not necessarily of the publisher.
Editorial Advisory Board
Herman Akdag, Université Paris 6, France
Slobodan Ribarić, University of Zagreb, Croatia
Reda Alhajj, University of Calgary, Canada
Nalin Sharda, Victoria University, Australia
Shi Kuo Chang, University of Pittsburgh, USA
Ling Shao, The University of Sheffield, UK
Alfredo Cuzzocrea, University of Calabria, Italy
Chia-Hung Yeh, National Sun Yat-Sen University, Taiwan
Chang-Tsun Li, University of Warwick, UK
Gang Zhang, Shenyang University of Technology, China
Table of Contents

Preface...................................................................................................................................................vii

Acknowledgment.................................................................................................................................. xii

Section 1

Chapter 1
Profiling User Color Perception for Image Retrieving............................................................................ 1
Imad El-Zakhem, Université de Reims Champagne Ardenne, France
Amine Aït-Younes, Université de Reims Champagne Ardenne, France
Herman Akdag, Université Paris 6, France
Hanna Greige, University of Balamand, Lebanon

Chapter 2
Approach of Using Texture and Shape for Image Retrieval.................................................................. 30
Gang Zhang, Northeastern University, China & Shenyang University of Technology, China
Zongmin Ma, Northeastern University, China
Li Yan, Northeastern University, China

Chapter 3
Mid-Level Image Descriptors................................................................................................................ 46
Jean Martinet, University of Lille, France
Ismail Elsayad, University of Lille, France

Chapter 4
An Image Retrieval Model Combining Ontology and Probabilistic Ranking....................................... 61
Lisa Fan, University of Regina, Canada
Botang Li, University of Regina, Canada

Chapter 5
Annotation of Medical Images............................................................................................................... 74
Chia-Hung Wei, Ching Yun University, Taiwan
Sherry Y. Chen, Brunel University, UK
Chapter 6
A Hybrid Approach to Content-Based Image Retrieval........................................................................ 91
Görkem Aşılıoğlu, TOBB ETÜ, Turkey
Emine Merve Kaya, TOBB ETÜ, Turkey
Duygu Sarıkaya, TOBB ETÜ, Turkey
Shang Gao, University of Calgary, Canada
Tansel Ozyer, TOBB ETÜ, Turkey
Jamal Jida, Lebanese University, Lebanon
Reda Alhajj, University of Calgary, Canada & Global University, Lebanon

Chapter 7
A Survey on Feature Based Image Retrieval Techniques.................................................................... 105
Ling Shao, University of Sheffield, UK

Chapter 8
Extracting Sport Video Semantics: Research and Applications........................................................... 121
Chia-Hung Yeh, National Sun Yat-sen University, Taiwan
Wen-Yu Tseng, National Sun Yat-sen University, Taiwan
Yu-Dun Lin, National Sun Yat-sen University, Taiwan
Chih-Chung Teng, National Sun Yat-sen University, Taiwan

Section 2

Chapter 9
Normal Forms for Multimedia Databases............................................................................................ 136
Shi Kuo Chang, University of Pittsburgh, USA
Vincenzo Deufemia, Università di Salerno, Italy
Giuseppe Polese, Università di Salerno, Italy

Chapter 10
Towards a Dynamic Semantic and Complex Relationship Modeling of Multimedia Data................. 154
Dawen Jia, Carleton University, Canada
Mengchi Liu, Carleton University, Canada

Chapter 11
Towards a Unified Multimedia Metadata Management Solution........................................................ 170
Samir Amir, University of Sciences and Technologies of Lille (USTL– Lille1), France
Ioan Marius Bilasco, University of Sciences and Technologies of Lille (USTL– Lille1), France
Md. Haidar Sharif, University of Sciences and Technologies of Lille (USTL– Lille1), France
Chabane Djeraba, University of Sciences and Technologies of Lille (USTL– Lille1), France
Chapter 12
Abstractions in Intelligent Multimedia Databases: Application of Layered Architecture
and Visual Keywords for Intelligent Search........................................................................................ 195
Ranjan Parekh, Jadavpur University, India
Nalin Sharda, Victoria University, Australia

Chapter 13
Fuzzy Logic for Image Retrieval and Image Databases: A Literature Overview................................ 221
Li Yan, Northeastern University, China
Z. M. Ma, Northeastern University, China

Chapter 14
EDUPMO: A Framework for Multimedia Production Management................................................... 239
Joni A. Amorim, Universidade Estadual de Campinas (UNICAMP), Brazil
Rosana G. S. Miskulin, Universidade Estadual Paulista “Júlio de Mesquita Filho” (UNESP), Brazil
Mauro S. Miskulin, Universidade Estadual de Campinas, Brazil

Chapter 15
Latent Semantic Analysis for Text Mining and Beyond...................................................................... 253
Anne Kao, Boeing Research & Technology, USA
Steve Poteet, Boeing Research & Technology, USA
Jason Wu, Boeing Research & Technology, USA
William Ferng, Boeing Research & Technology, USA
Rod Tjoelker, Boeing Research & Technology, USA
Lesley Quach, Boeing Research & Technology, USA

Compilation of References................................................................................................................ 281

About the Contributors..................................................................................................................... 307

Index.................................................................................................................................................... 316
 vii

Preface

The decreasing costs of consumer electronic devices such as digital cameras and digital camcorders,
along with the ease of transportation facilitated by the Internet, has lead to a phenomenal rise in the
amount of multimedia data. Now multimedia data comprising of images, audio, and video is becoming
increasingly common. Given that this trend of increased use of multimedia data is likely to accelerate,
there is an urgent need for providing a clear means of capturing, storing, indexing, retrieving, analyzing,
and summarizing such data.
Image data, for example, is a very commonly used multimedia data. The early image retrieval sys-
tems are based on manually annotated descriptions, called text-based image retrieval (TBIR). TBIR
is a great leap forward, but has several inherent drawbacks. First, textual description is not capable of
capturing the visual contents of an image accurately, and in many circumstances, textual annotations are
not available. Second, different people may describe the content of an image in different ways, which
limits the recall performance of textual-based image retrieval systems. Third, for some images there is
something that no words can convey. To resolve these problems, content-based image retrieval (CBIR)
systems are designed to support image retrieval, and have been used since the early 1990s. Also, some
novel approaches (e.g., relevance feedback, semantic understanding, semantic annotation, and semantic
retrieval of images) have been developed in the last decade to improve image retrieval and satisfy the
advanced requirements of image retrieval.
Multimedia data retrieval is closely related to multimedia data management. Multimedia data
management facilitates the manipulation of multimedia data such as representation, storage, index, re-
trieval, maintenance, and so on. Multimedia data retrieval is the key to implementing multimedia data
management on one hand. On the other hand, multimedia data retrieval should be carried out based
on multimedia data representation, storage, and index, which are the major tasks of multimedia data
management. Databases are designed to support the data storage, processing, and retrieval activities
related to data management, and database management systems can provide efficient task support and
tremendous gain in productivity is hereby accomplished using these technologies. There is no doubt that
database systems play an important role in multimedia data management, and multimedia data manage-
ment requires database technique support. Multimedia databases, which have become the repositories
of large volumes of multimedia data, are emerging.
Multimedia databases play a crucial role in multimedia data management, which provide the
mechanisms for storing and retrieving multimedia data efficiently and naturally. Being a special kind
of databases, multimedia databases have been developed and used in many application fields. Many
researchers have been concentrating on multimedia data management using multimedia databases. The
research and development of multimedia data management using multimedia databases are receiving
increasing attention. By means of multimedia databases, large volumes of multimedia data can be stored
viii

and indexed and then retrieved effectively and naturally from multimedia databases. Intelligent multi-
media data retrieval systems are built based on multimedia databases to support various problem solving
and decision making. Thus, intelligent multimedia databases and information retrieval is a field that
must be investigated by academic researchers together with developers both from CBIR and AI fields.
The book has two focuses on multimedia data retrieval and multimedia databases, aiming at provid-
ing a single account of technologies and practices in multimedia data management. The objective of the
book is to provide the state of the art information to academics, researchers, and industry practitioners
who are involved or interested in the study, use, design, and development of advanced and emerging
multimedia data retrieval and management with ultimate aim to empower individuals and organizations
in building competencies for exploiting the opportunities of the knowledge society. This book presents
the latest research and application results in multimedia data retrieval and management. The different
chapters in the book have been contributed by different authors and provide possible solutions for the
different types of technological problems concerning multimedia data retrieval and management.

INTRODUCTION

This book, which consists of fifteen chapters, is organized into two major sections. The first section
discusses the feature and semantics of multimedia data as well as their usage in multimedia information
retrieval in the first eight chapters. The next seven chapters covering database and intelligence technolo-
gies for multimedia data management comprise the second section.
First of all, we take a look at the issues of the feature and semantics of multimedia data as well as
their usage in multimedia information retrieval.
Imad EL-Zakhem et al. concentrate on building a user profile according to his own perception of
colors for image retrieving. They develop a dynamic construction of the user profile, which will increase
their satisfaction by being more personalized and accommodated to their particular needs. They suggest
two methods to define the perception and transform it into a profile: the first one is achieved by querying
the user and getting answers and the second one is achieved by comparing different subjects and ending
up by an appropriate aggregation. They also present a method recalculating the amount of colors in the
image based on another set of parameters, and the colorimetric profile of the image is being modified
accordingly. Avoiding the repetition of the process at the pixel level is the main target of this phase,
because reprocessing each image is time consuming and not feasible.
In content-based image retrieval, different kinds of features (e.g., texture features, color features and
shape features) may be used jointly, and feature integration is hereby one of crucial issues in content-
based image retrieval. Gang Zhang et al. develop an approach of integrating shape and texture features
and investigate if integration features are more discriminative than single features. Single feature ex-
traction and description is foundation of the feature integration. They apply Gabor wavelet transform
with minimum information redundancy to extract texture features, which are used for feature analyses.
Fourier descriptor approach with brightness is used to extract shape features. Then both features are
integrated together by weights. They make the comparisons among the integration features, the texture
features, and the shape features so that the discrimination of the integration features can be testified.
The research domain of automatic image annotation and search from low-level descriptors analy-
sis has considerably evolved in the last 15 years. Since then, this domain has reached a level of ma-
turity where only small improvements are brought in new models and systems. Jean Martinet and
Ismail Elsayad propose a classification of image descriptors, from low-level descriptors to high-level
 ix

descriptors, introducing the notion of mid-level descriptors for image representation. A mid-level de-
scriptor is described as an intermediate representation between low-level descriptors (derived from the
signal) and high-level descriptors (conveying semantics associated to the image). Mid-level descriptors
are built for the purpose of yielding a finer representation for a particular set of documents. They describe
a number of image representation techniques from a mid-level description perspective.
There are hundreds of millions of images available on the current World Wide Web, and the demand
for image retrieval and browsing online is growing dramatically. The typical keyword-based retrieval
methods for multimedia documents assume that the user has an exact goal in mind in searching a set of
images whereas users normally do not know what they want, or the user faces a repository of images
whose domain is less known and content is semantically complicated. In these cases it is difficult to
decide what keywords to use for the query. Lisa Fan and Botang Li present an approach of the user-
driven ontology guided image retrieval. It combines (a) the certain reasoning techniques based on logic
inside ontology and (b) the uncertain reasoning technique based on Bayesian Network to provide users
the enhanced image retrieval on the Web. Their approach is for easily plugging in an external ontology
in the distributed environment and assists user searching for a set of images effectively. In addition, to
obtain a faster real-time search result, the ontology query and BN computation should be run on the
off-line mode, and the results should be stored into the indexing record.
A large number of digital medical images have been produced in hospitals in the last decade. These
medical images are stored in large-scale image databases and can facilitate medical doctors, profession-
als, researchers, and college students to diagnose current patients and provide valuable information for
their studies. Image annotation is considered as a vital task for searching, and indexing large collec-
tions of medical images. Chia-Hung Wei and Sherry Y Chen present a complete scheme for automatic
annotation on mammograms. Firstly, they present the feature extraction methods based on BI-RADS
standards. This ensures that the meaning and interpretation of mammograms are clearly characterized
and can be reliably used for feature extraction. Secondly, they propose the SVM classification approach
to image annotation. Finally, their experimental results demonstrate that the scheme can achieve fair
performance on image annotation.
Digital image storage and retrieval is gaining more popularity due to the rapidly advancing technology
and the large number of vital applications, in addition to flexibility in managing personal collections of
images. Traditional approaches employ keyword based indexing which is not very effective. Content
based methods are more attractive though challenging and require considerable effort for automated
feature extraction. Görkem Aşılıoğlu et al. present a hybrid method for extracting features from images
using a combination of already established methods, allowing them to be compared to a given input
image as seen in other query-by-example methods. First, the image features are calculated using edge
orientation autocorrelograms and color correlograms. Then, distances of the images to the original im-
age are calculated using the L1 distance feature separately for both features. The distance sets are then
be merged according to a weight supplied by the user.
Disadvantages with text-based image retrieval have provoked growing interest in the development
of Content-Based Image Retrieval (CBIR). In CBIR, instead of being manually annotated by text-based
keywords, images are indexed by their visual content, such as color, texture, etc. Ling Shao surveys
content-based image retrieval techniques on representing and extracting visual features, such as color,
shape, and texture. The feature representation and extraction approaches are first classified and discussed.
Then, he summarizes several classical CBIR systems which rely on either global features or features
detected on segmented regions. The inefficiency and disadvantages of those narrow-domain systems
x

are also presented. Finally, he discusses two recent trends on image retrieval, namely semantic based
methods and local invariant regions based methods, and proposes directions for future work.
With the rapid growth of digital videos, efficient tools are essential to facilitate content indexing,
searching, retrieving, browsing, skimming, and summarization. Sport video analysis aims to identify
what excites audiences. Previous methods rely mainly on video decomposition, using domain specific
knowledge and lacking the ability to produce personalized semantics especially in highlight detection.
Research on suitable and efficient techniques for sport video analysis has been conducted extensively
over the last decade. Chia-Hung Yeh et al. review the development of sport video analysis and explore
solutions to the challenge of extracting high-level semantics in sport videos. They propose a method
to analyze baseball videos via the concept of gap length. Use-interaction may be a solution to achieve
personalization in semantics extraction. The techniques introduced can be wildly applied to many fields,
such as indexing, searching, retrieving, summarization, skimming, training, and entertainment.
The second section deals with the issues of database and intelligence technologies for multimedia
data management.
The last decades have witnessed a considerable rise in the amount of multimedia data. Data models
and database management systems (DBMSs) can play a crucial role in the storage and management of
multimedia data. Being a special kind of database systems, multimedia databases have been developed
and used in many application fields. Shi Kuo Chang, Vincenzo Deufemia, and Giuseppe Polese present
normal forms for the design of multimedia database schemes with reduced manipulation anomalies.
They first discuss how to describe the semantics of multimedia attributes based upon the concept of
generalized icons, already used in the modeling of multimedia languages. They then introduce new ex-
tended dependencies involving different types of multimedia data. Based upon these new dependencies,
they define five normal forms for multimedia databases, some focusing on the level of segmentation of
multimedia attributes, others on the level of fragmentation of tables. Thus a normalization framework
for multimedia databases is developed, which provides proper design guidelines to improve the quality
of multimedia database schemes.
Multimedia data is a challenge for data management. The semantics of traditional alphanumeric
data are mostly explicit, unique, and self-contained, but the semantics of multimedia data are usually
dynamic, diversiform, and varying from one user’s perspective to another’s. Dawen Jia and Mengchi
Liu introduce a new model, titled the Information Networking Model (INM). It provides a strong se-
mantic modeling mechanism that allows modeling of the real world in a natural and direct way. With
INM, users can model multimedia data, which consists of dynamic semantics. The context-dependency
and media-independency features of multimedia data can easily be represented by INM. In addition,
multimedia multiple classifications are naturally supported. Based on INM, they propose a multimedia
data modeling mechanism in which users can take advantage of basic multimedia metadata, semantic
relationships, and contextual semantic information to search multimedia data.
With increasing use of multimedia in various domains, several metadata standards appeared these
last decades in order to facilitate the manipulation of multimedia contents. These standards help con-
sumers to search content they desire and to adapt the retrieved content according to consumers’ pro-
files and preferences. However, in order to extract information from a given standard, user must have
a pre-knowledge about this latest. This condition is not easy to satisfy due to the increasing number
of available standards. Samir Amir et al. first give an overview about existing multimedia metadata
standards and CAM4Home project initiative that covers a wide area of information related to multime-
dia delivery and includes multimedia content description, user preference and profile description, and
devices’ characteristic description. Then they relate about multimedia and generic integration issues
 xi

by discussing the work done by W3C working group in order to integrate heterogeneous metadata and
some generic approaches providing mapping between ontologies. They also consecrate to the illustra-
tion of the proposal of a new architecture for the multimedia metadata integration system and discuss
about challenges of its realization.
Semantic characterization is necessary for developing intelligent multimedia databases, because hu-
mans tend to search for media content based on their inherent semantics. However, automated inference
of semantic concepts derived from media components stored in a database is still a challenge. Ranjan
Parekh and Nalin Sharda demonstrate how layered architectures and visual keywords can be used to
develop intelligent search systems for multimedia databases. The layered architecture is used to extract
meta-data from multimedia components at various layers of abstractions. To access the various abstracted
features, a query schema is presented which provides a single point of access while establishing hierarchi-
cal pathways between feature-classes. Minimization of the semantic gap is addressed using the concept
of visual keyword (VK). Semantic information is however predominantly expressed in textual form,
and hence is susceptible to the limitations of textual descriptors–viz. ambiguities related to synonyms,
homonyms, hypernyms, and hyponyms. To handle such ambiguities they propose a domain specific
ontology-based layer on top of the semantic layer, to increase the effectiveness of the search process.
Fuzzy set theory has been extensively applied to the representation and processing of imprecise and
uncertain data. Image data is becoming a kind of important data resources with rapid growth in the num-
ber of large-scale image repositories. But image data is fuzzy in nature and imprecision and vagueness
may exist in both image descriptions and query specifications. Li Yan and Z. M. Ma review some major
work of image retrieval with fuzzy logic in the literature, including fuzzy content-based image retrieval
and database support for fuzzy image retrieval. For the fuzzy content-based image retrieval, they present
how fuzzy sets are applied for the extraction and representation of visual (colors, shapes, textures) fea-
tures, similarity measures and indexing, relevance feedback, and retrieval systems. For the fuzzy image
database retrieval, they present how fuzzy sets are applied for fuzzy image query processing based on a
defined database models, and how various fuzzy database models can support image data management.
Project portfolio management of multimedia production and use emerges today as a challenge both for
the enrichment of traditional classroom based teaching and for distance education offering. In this way,
Joni A. Amorim, Rosana G. S. Miskulin, and Mauro S. Miskulin intend to answer the following ques-
tion: “Which are the fundamental aspects to be considered in the management of projects on educational
multimedia production and use?” They present a proposal of a project management model for digital
content production and use. The model, the methodology and the implementation are named EduPMO
(Educational Project Management Office). The model, the methodology and the implementation should
be understood as related but independent entities. This interdisciplinary investigation involves differ-
ent topics, going from metadata and interoperability to intellectual property and process improvement.
Latent Semantic Analysis (LSA) or Latent Semantic Indexing (LSI), when applied to information
retrieval, has been a major analysis approach in text mining. It is an extension of the vector space method
in information retrieval, representing documents as numerical vectors, but using a more sophisticated
mathematical approach to characterize the essential features of the documents and reduce the number of
features in the search space. Anne Kao et al. summarize several major approaches to this dimensionality
reduction, each of which has strengths and weaknesses, and describe recent breakthroughs and advances.
They show how the constructs and products of LSA applications can be made user-interpretable and review
applications of LSA beyond information retrieval, in particular, to text information visualization. While
the major application of LSA is for text mining, it is also highly applicable to cross-language information
retrieval, Web mining, and analysis of text transcribed from speech and textual information in video.
xii

Acknowledgment

The editors wish to thank all of the authors for their insights and excellent contributions to this book and
would like to acknowledge the help of all involved in the collation and review process of the book,
without whose support the project could not have been satisfactorily completed. Most of the authors of
chapters included in this book also served as referees for chapters written by other authors. Thanks go
to all those who provided constructive and comprehensive reviews.
A further special note of thanks goes to all the staff at IGI Global, whose contributions throughout
the whole process from inception of the initial idea to final publication have been invaluable. Special
thanks also go to the publishing team at IGI Global. This book would not have been possible without
the ongoing professional support from IGI Global.
The idea of editing this volume stems from the initial research work that the editors did in past sev-
eral years. The research work of the editors was supported by the Program for New Century Excellent
Talents in University (NCET-05-0288).

Li Yan
Northeastern University, China

Zongmin Ma
Northeastern University, China
Section 1
 1

Chapter 1
Profiling User Color Perception
for Image Retrieving
Imad El-Zakhem
Université de Reims Champagne Ardenne, France

Amine Aït-Younes
Université de Reims Champagne Ardenne, France

Herman Akdag
Université Paris 6, France

Hanna Greige
University of Balamand, Lebanon

ABSTRACT
The aim of this work is to build a user profile according to his own perception of colors for image retriev-
ing. Images are being processed relying on a standard or initial set of parameters using the fuzzy set
theory and the HLS color space (Hue, Lightness, and Saturation). We developed a dynamic construction
of the user profile, which will increase his satisfaction by being more personalized and accommodated to
his particular needs. We suggest two methods to define the perception and transform it into a profile; the
first method is achieved by querying the user and getting answers, which will guide through the process
of implementation of the profile; the second method is achieved by comparing different subjects and
ending up by an appropriate aggregation. We also present a method that will recalculate the amount of
colors in the image based on another set of parameters, so the colorimetric profile of the image is being
modified accordingly. Avoiding the repetition of the process at the pixel level is the main target of this
phase, because reprocessing each image is time consuming and turned to be not feasible.

DOI: 10.4018/978-1-61350-126-9.ch001

Copyright © 2012, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global is prohibited.
 Profiling User Color Perception for Image Retrieving

INTRODUCTION The user’s relevance feedbacks are modeled by

fuzzy sets, i.e. the user expresses his satisfaction
Image retrieving is an important problem that or discontentment by assigning a label to the
can be useful in many fields (Foulloy, 1990), retrieved images. A dissimilarity based on fuzzy
Hammami (2002), Hong (2000) and Le Saux integrals is then used. It is a kind of supervised
(2003)). For example, in medical applications, it learning for image retrieval systems.
is important to retrieve images in order to help All the aforementioned authors work with a
medical expert forecasts. Another example lies in query image. That is not the case of Binaghi et al
web content detection: classification of images who use a user query expressed by crisp values
to determine whether they contain a lot of skin of colors (Binaghi, 1994). More precisely, they
texture or not in order to detect adult and sexual provide methodological and technical solutions
contents (Hammami, 2002). to compute similarities between the query and
There are several works on image classification the image index. The user also has to choose the
based on the determination of a similarity degree color dimension (hue, chroma or lightness), the
between images. This kind of classification can image area covered by the referent color and the
be done through several techniques, for example: type of color distribution in the image. Thus, the
statistical approach like Support Vector Machines user has to know exactly what he is looking for
(Hong, 2000; Barla, 2003; Vapnik, 1998), color and in terms of colors and color distribution in order
illumination features using histograms intersec- to obtain satisfactory results. That is why we have
tion (Barla, 2003; Bourghorbel, 2002) and fuzzy focused on the problem of the query expression
logic (Chen, 2002; Omhover, 2004; Vertan, 2000). which is very simple in our case: the user can ask
Among these Image Retrieval Systems, we only for a certain tone if he wants.
distinguish at least two kinds: those that consider The aim of this work is not to make a classical
the histograms and those that don’t. Barla et al classification but to retrieve images according
address the problem of classifying images by to their dominant(s) color(s) expressed through
exploiting color and illumination features, using linguistic expressions. In this work, Images are
histogram intersections. The histogram intersec- processed using a fuzzy representation of colors
tion is used as a kernel function for SVMs and based on the HLS space. The image processing
allows one to classify images by similarity of consists of modeling the three dimensions of color
histograms (Barla, 2003). (hue, saturation and lightness) by using fuzzy
Han and Ma propose a fuzzy color histogram membership functions.
that permits to consider the color similarity across The standard colorimetric profile of each image
different bins and the color dissimilarity in the is build using standard values of the membership
same bin (Han, 2002). Thus, as in Vertan (2000), functions. These profiles may not be accepted by
a pixel of a given color will contribute not only all users since the perception is a subjective issue.
to its specific bin, but its membership value will To resolve this problem, the user is asked to build
be spread to other histogram bins. his profile, thus when retrieving images, they will
Another kind of approach is presented by Wang be brought regarding his perception.
and Du: they propose an algorithm for indexing To avoid the reprocessing of images, we use
and retrieving images based on region segmenta- a new approach by applying a transformation
tion, and they also compute similarities between procedure on the standard colorimetric profile of
images in order to classify them (Wang, 2000). the images according to user’s perception. In this
As for Frigui, he describes a system that offers procedure, we use the notions of comparability
the refinement of the user query (Frigui, 2001). and compatibility of fuzzy subsets

2
Profiling User Color Perception for Image Retrieving

Figure 1. An example of a color histogram

variables defined on the chromatic circle, may
allow a better handling of the color parameters
for quantitative purposes. Many research studies
took interest in the polar space color representa-
tions (Kender, 1976; Smith, 1978; Shih, 1995;
Carron, 1995).
In many color-based image retrieval{XE “re-
trieval”} applications (CBIR), the color properties
of an image are characterized by the probability
distribution of the colors in the image. Based on the
color space representation, many CBIR systems
represent region colors by means of histograms, so
that they can be stored in a simple, intuitive way.
Histograms may be represented in two-dimen-
sional or three-dimensional spaces (Figure 1).
The color histogram remains the most popular
The chapter is organized as follows: after a
representation of color distributions since it is
discussion about the color representation we pres-
insensitive to small object distortions and easy to
ent in section 2 the problem of color representa-
compute. However, it is not very efficient due to
tion with fuzzy membership functions and image
its large memory requirement. Typically, a color
processing. The global approach for building a
histogram might consist of 512 bins. With such
user profile is presented in section 3. In section 4,
a large number of bins, the performance of current
the notions of comparability and compatibility of
indexing techniques is reduced to a sequential
fuzzy subsets are presented. Finally, some statisti-
scanning (Webe, 1998; Rui, 1999). To make
cal results are presented in section 5.
color histogram-based image retrieval truly scal-
able to large image databases it is desirable to
reduce the number of parameters needed to de-
COLOR REPRESENTATION
scribe the histogram while still preserving the
retrieval performance.
Colors are being represented through different
Approaches to deal with these problems include
color spaces like the RGB, HLS, Lab etc.
the usage of coarser histograms (Mitra, 1997;
The CIE (Commission Internationale de
Pass, 1999), dominant colors or signature colors
l’Eclairage) has standardized the weights of the
(Androutsos, 1999; Deng, 2001) and application
triplet R, G,B. The intensities of the three com-
of signal processing compression techniques such
ponents are additive, so that all colors accessible
as the Karhunen–Loève transform, discrete cosine
from an RGB basis are obtained by sums of the
transform, Hadamard transform, Haar transform,
primary colors R, G, and B and by multiplications
and wavelets, etc. (Albuz, 2001; Berens, 1995;
by nonnegative constants (CIE, 1986).
Hafner, 1995).
Despite its use in computer systems, the
Also another drawback of histograms is that
RGB color representation has some drawbacks
they tend to be noisy. Sometimes it is necessary to
like the poor human interpretation, the none-
extract only relevant information from complete
uniformity, etc.
histograms, thus retaining reduced feature vec-
The polar representation, like the HLS and
tors. Some authors rely on extracting their Color
HSV, uses one variable for light intensity and two

3
 Profiling User Color Perception for Image Retrieving

Figure 2. The RGB space

Moments or Color Sets (Diplaros, 2006; Swain, distribution of the pixels on the three axes (red,
1991; Li, 1999). green, blue). However with this kind of histogram
Color moments are good descriptors when the it is difficult to define a fuzzy membership degree
analyzed region presents a dominant color, but this to a given color, for example how to define in a
technique provides confusing results when applied fuzzy way a “pink”?
to mixed color regions. Color Sets can be adjusted The RGB space is not appropriate for our
to represent color regions, but they introduce a problem also because three dimensions (R, G and
quantization on the original color histograms, B) are necessary to identify a color. To facilitate
thus, they do not adapt to the characteristics of the color identification we choose another space
any particular colored region. that allows us to characterize a color with only
one dimension: its hue. Indeed hue is enough to
recognize the color, except when the color is very
FUZZY REPRESENTATION OF pale or very somber. This other space is called
COLORS AND IMAGE PROCESSING HLS (Hue, Lightness, and Saturation) where
lightness corresponds to the quantity of “white”
As we know, one of the spaces usually used to in the color and saturation corresponds to the light
represent the color on a screen is the RGB space intensity of the color. The identification of color
(Red, Green and Blue). It is a three dimensional is made in two steps: first H, then L, S. The HLS
space representing the three primary colors that space can be represented through a cylinder or a
usually vary from 0 to 255. The origin of this bi-cone (Figure 3).
space (0, 0, 0) corresponds to the lack of color Moreover, in practice, various models of
which represents the “black” color. On the other color representation use also a “two-step” iden-
hand the point (255,255,255) corresponds to the tification of color. For example, Aron Sigfrid
maximum of color which represents the “white”. Forsius, Pantone Matching System, RAL (Reich-
The representation of the colors in this space gives sAusschuss, für Lieferbedingungen und Gütesi-
us a cube (Figure 2). cherung), Munsel, ISCC-NBS (Inter-Society
In classification methods, this space is used Color Council - National Bureau of Standards),
in the calculation of similarity through color etc. (cf. Couwenbergh(2003)) use at first a color
histogram (Swain, 1991) which represents the

4
Profiling User Color Perception for Image Retrieving

Figure 3. The HLS space

description by means of the hue then a refinement (h = 85) or blue (h = 170) while they perceive it
through the saturation and the lightness. very well with orange (h = 21) for example.
H is defined as an angle but we can also To deal with non uniformly distributed scales,
represent it in the interval [0,255] as the other authors such as Herrera and Martinez propose to
components L and S. The difference between H use fuzzy linguistic hierarchies with more or less
and the other components is that its definition labels, depending on the desired granularity (Her-
interval loops which means that 0 and 256 are rera, 2001). Another approach from Truck (2001)
the same points. The “pure” red (represented in is to represent the hues with trapezoidal or trian-
RGB space by the point (255, 0, 0)) corresponds gular fuzzy subsets thanks to colors definitions
to an angle equal to 0° for h, a saturation s equal from www.pourpre.com. This technique is more
to 255 and a lightness l equal to 128. empirical but fits better the human perception,
For this problem, we limit ourselves to the that is why we also use this approach.
nine fundamental colors defined by the set For each color of  they built a membership
 representing a good sample of colors function varying from 0 to 1 (ft with t ∈  ). If
(dimension H): this function is equal to 1, the corresponding
color is a “true color” (cf. Figure 4).
 = {red, orange, yellow, green, cyan, blue, These functions were built using colors defini-
purple, magenta, pink} tion (www.poupre.com). For each fundamental
color, the associated interval is defined according
This set corresponds to the seven colors of to linguistic names of colors. For example to
Newton (cf. Roire(2000)) to which we have added construct fyellow, we can use color “mustard” whose
color pink and color cyan. Of course, this choice hue is equal to 55 and whose membership to fyellow
is not restrictive; we can modify the set of colors is equal to 0.5.
as desired. For some colors, the result gives a wide interval.
It is the case for the colors “green” and “blue”
Fuzzy Representation of Colors which are represented by trapezoidal fuzzy subsets.
For the construction of these functions, in this
As we have seen HLS space is convenient for our article we suppose that two functions representing
problem but it is a non UCS (uniform color scale) two successive colors have their intersection point
space (Herrera, 2001). Indeed our eyes don’t per- value equal to 1/2. It means that when h corre-
ceive small variations of hue when color is green sponds to an intersection point it can be assigned
to both colors with the same weight.

5
 Profiling User Color Perception for Image Retrieving

Figure 4. The dimension H

Figure 5. Trapezoidal fuzzy subset

As usual (Bouchon-Meunier, 1995) we denote 


(a, b, α, β) a trapezoidal fuzzy subset (Figure 5).  0 if h ≥ 43
 h
When the kernel is reduced to only one point, it is forange (h ) =  if h < 21 (2)
a triangular subset denoted by (a, α, β) since a=b.  21

Now we can define the membership function  43 − h if h ≥ 21
of any color t:  22

 1 if h ≥a For t = green we have a trapezoidal subset


 ∧ h ≤b with (a = 75, α = 22, b = 95, β = 33):

 0 if h ≤ a − α
  1 if h ≥ 75
 ∧ h ≥b +β
h − (a − α) 
 ∧ h ≤ 95
∀t ∈ T , ft (h ) =  if h >a −α  0
 α  if h ≤ 43
 
 ∧ h <a ∧ h ≥ 128
(b + β)) − h 
 if h >b fgreen (h ) =  h − 43 if h > 43 (3)
 β  22
 
 ∧ h <b +β  ∧ h < 75

(1) 128 − h
 333 if h > 95

For example, for t = orange we have a trian-  ∧ h < 128
gular subset with:

6
Profiling User Color Perception for Image Retrieving

Figure 6. Fundamental color qualifiers

Moreover if we want to complete the model- The membership function of any qualifier q
ing, it is necessary to take into account the two is defined below:
other dimensions (L, S). A scale representing the
colorimetric qualifiers is associated to each dimen-  1 if a ≤s ≤b

 ∧ c ≤l ≤d
sion. These two intervals are divided into three: 
the first subinterval corresponds to a low value,  0 if s ≤ a − α ∨ s ≥ b + β

∨ l ≤c −γ ∨l ≥d +δ
the second to an average value and the last to a 
 l − (c − γ )
strong value. This division gives for saturation S: 

if c −γ <l <c
γ
“dull”, “moderately dull” and “saturated”; and 
 ∧ αl − γs ≤ αc − γa
for lightness L: “gloomy”, “heavy” and “pallid”. 
∧ βl + γs ≤ βc + γb

These two scales are then aggregated to give  (d + δ) − l
 if d <l <d +δ
nine qualifiers for colors defined by the following  δ
∀q ∈ , fq (l , s ) = 
set (Figure 6) (Aït Younes, 2007):  ∧ βl − δs > βd − δb

 ∧ αl + δs > αd + δa

 = {somber, dark, deep, gray, medium, bright, s − (a − α) if a −α <s <a

pale, light, luminous}. 
α
 ∧ αl − γs > αc − γa

 ∧ αl + δs ≤ αd + δa
Each element of the set  is associated to a (b + β) − s
membership function varying between 0 and 1 ( 
 β
if b <s <b +β
fq withq ∈  ). For these functions the intersec- 
 ∧ βl + γs > βc + γb

tion point value is also supposed equal to 1/2 (cf.  ∧ βl − δs ≤ βd − δb
Figure 8)). Every function is represented through (4)
the set (a, b, c, d, α, β, γ, δ) (Figure 7).

7
 Profiling User Color Perception for Image Retrieving

Figure 7. Trapezoidal 3-D fuzzy subset

Figure 8. Dimensions L and S

8
Profiling User Color Perception for Image Retrieving

For example, for q = somber, we have (a = which are associated to fuzzy membership func-
α = 0, b = 43, β = 84, c = γ = 0, d = 43, δ = 84): tions: fdark , fmedium and flight (Figure 9).
For example, for t= black we have (a = α = 0,
 1 if s ≤ 43 b = 255, β = 0, c = γ = 0, d = 15, δ = 10):

 ∧ l ≤ 43

 0 if s ≥ 127 

  1 if l ≤ 15
 ∨ l ≥ 127
 fblack (l ) =  0 if l ≥ 25 (6)
fsomber (l , s ) =  127 − l

if 43 < l < 127 
 84  25 − l
  if 15 < l < 25
 ∧ l >s  10

127 − s if 43 < s < 127
 84
 Image Processing
 ∧ l ≤s
(5)
Each pixel can now be categorized in one or more
We also took into account the colors black, fuzzy classes with an associated membership
gray and white. Fuzzy membership functions are degree. Thus, it is possible to compute the mem-
associated to this colors (fblack, , fwhite and fgray). bership functions to the various classes (colors
These colors are completely defined through the and qualifiers) for the whole image.
spaces L and S. If the lightness is very low then Let I be an image and  be the set represent-
the color is white, and a high level of the lightness ing the pixels of I.
means the color is black. If the saturation is very Each element p of the set  is defined by its
low then the color is gray. For this last color we color coordinates (hp, lp, sp). We calculate the
define three qualifiers: dark, medium and light f u n c t i o n s f t ( h p ) , fq (l p , s p ) , fq (l p , s p ) f o r
t ∈ T and q ∈ Q.

Figure 9. Black, gray and white

9
 Profiling User Color Perception for Image Retrieving

Let Ft and Ft ,q be the following functions, Thus, an image can be assigned to several class-
representing the membership degree of I to the es, and it can be assigned to a subclass only if it
classes t and (t,q): is also assigned to its father class. For example,
an image cannot be assigned to “red, bright” class
∑ f (h )
t p
(Cred ,bright ) if it is not assigned to the “red” one
•∀t ∈ T , Ft (I ) = p ∈P
(Cred).
|P| The standard colorimetric profile of each
image is build using the standard values of the
∑ f (l , s
q p p ) × gt (hp ) membership functions as described above. These
•∀(t, q ) ∈ T ×Q, Ft ,q (I ) = p ∈P profiles may not be accepted by all users since the
|P| perception is a subjective issue. To resolve this
0 if ft (hp ) = 0 problem, the user is asked to build his profile,
With gt (hp ) = 
1 else thus when retrieving images, they will be brought
regarding his perception.
The profile associated to the image is denoted
by [Ft(I), Ft ,q (I )] and contains 96 elements: | 
| + | T ×Q |+ | {black, white, gray} | + | {gray} × USER PROFILE FOR
{dark, medium, light} |= 9 + 81 + 3 + 3 (Figure COLOR PERCEPTION
10).
Because of the representation of colors in two The aim is to build for each user his own member-
steps (hues on the one hand, and qualifiers on the ship functions of colors and qualifiers (Figure 4
other hand), the classes follow a hierarchy: and Figure 8). Starting from the initial functions
classes Ctwitht ∈  ∪ {black , white, gray } can defined in the previous section, we will modify
be considered as fathers and the classes the function parameters for each user.
Cc,q with (c, q ) ∈ T ∪ {gray } ×Q as their sons. The user is invited to build his profile by
Let us denote: identifying himself as an expert or as a simple
user. An expert user will follow a many-steps
F * (I ) = maxt ∈ (Ft (I )) algorithm; on the other hand the task for a simple
user is much more simplified. The experts profiles
are built using a questionnaire while non-experts’
iFt * (I ) = maxq ∈Q (Ft ,q (I )) ∀t ∈ T , by comparison of subjects (El-Zakhem, 2008).
and for t = gray, q ∈ {dark, medium, light }
Expert Algorithm

An image I will be assigned to: We mean by expert, the user who is interested to
build a detailed colorimetric profile and who is
• the classes Ct if Ft (I ) ≥ F * (I ) − λ , intended to respond to a questionnaire R.
For each color i in the set T, we denote its
∀t ∈  ∪ {black , white, gray } with λ a
previous by (i-1) and its successive by (i+1); also
tolerance threshold.
each color i is defined by a value mi in the set
• the classes Ct ,q if Ft (I ) ≥ F * (I ) − λ and ,
 . Each value in  corresponds to a color in
T (the middle of the kernel of the membership
∀(t, q ) ∈ T ×Q ∪ {gray } × {dark, medium, light }
function (Figure 4)).

10
 Another Random Document on
Scribd Without Any Related Topics
The Project Gutenberg eBook of The Pyrotechnist's
Treasury; Or, Complete Art of Making Fireworks
This ebook is for the use of anyone anywhere in the United States and
most other parts of the world at no cost and with almost no restrictions
whatsoever. You may copy it, give it away or re-use it under the terms of
the Project Gutenberg License included with this ebook or online at
www.gutenberg.org. If you are not located in the United States, you will
have to check the laws of the country where you are located before using
this eBook.

Title: The Pyrotechnist's Treasury; Or, Complete Art of Making Fireworks

Author: Thomas Kentish

Release date: September 29, 2012 [eBook #40892]

Most recently updated: October 23, 2024

Language: English

Credits: Produced by Paul Clark and the Online Distributed

Proofreading Team at http://www.pgdp.net (This book was
produced from scanned images of public domain material
from the Google Print project.)

*** START OF THE PROJECT GUTENBERG EBOOK THE PYROTECHNIST'S

TREASURY; OR, COMPLETE ART OF MAKING FIREWORKS ***
Transcriber's Note:
Every effort has been made to replicate this text as faithfully as possible, including some inconsistencies
in hyphenation. Some changes of spelling and punctuation have been made. They are listed at the end
of the text.

THE PYROTECHNIST'S TREASURY.

 "The Secret Out Series."
Crown 8vo, cloth extra, profusely Illustrated, price 4s. 6d. each.

Art of Amusing. A Collection of Graceful Arts, Games, Tricks, Puzzles, and

Charades, intended to amuse everybody. By Frank Bellew. With nearly 300
Illustrations.
Hanky-Panky. A Wonderful Book of Very Easy Tricks, Very Difficult Tricks,
White Magic, Sleight of Hand; in fact, all those startling Deceptions which
the Great Wizards call "Hanky-Panky." Edited by W. H. Cremer. With nearly
200 Illustrations.
Magician's Own Book. Ample Instructions for Performances with Cups and
Balls, Eggs, Hats, Handkerchiefs, &c. All from actual experience. Edited by
W. H. Cremer. With 200 Illustrations.
Magic No Mystery: Tricks with Cards, Dice, Balls, &c., with fully descriptive
working directions; How to train Performing Animals; the art of Secret
Writing, &c. Numerous Illustrations.
Merry Circle (The), and How the Visitors were Entertained during Twelve
Pleasant Evenings. A Book of New Intellectual Games and Amusements.
Edited by Mrs. Clara Bellew. With numerous Illustrations.
Secret Out; or, One Thousand Tricks with Cards, and other Recreations;
with Entertaining Experiments in Drawing Room or "White Magic." Edited
by W. H. Cremer. With 300 Engravings.

⁂ The above Series of Books contain Explanations of all the most startling
feats of legerdemain performed by the celebrated Prestidigitateurs, Robert
Houdin, Frikell, Dr. Lynn, Professor Anderson, and others.
CHATTO & WINDUS, Piccadilly, W.

THE PYROTECHNIST'S TREASURY;

OR,
 COMPLETE ART OF MAKING FIREWORKS.
BY
THOMAS KENTISH.

Flammas Jovis, et sonitus imitatur Olympi.

Virgil, Æn. vi. 586.
London: CHATTO & WINDUS, PICCADILLY. 1878.
[All rights reserved.]

LONDON:
PRINTED BY J. OGDEN AND CO.,
172, ST. JOHN STREET, E.C.
 PREFACE.
Until within the last few years there was scarcely any work in English, that I am
aware of, on the subject of Pyrotechny, worth reading, with the exception of an
Article in "Brewster's Cyclopædia," by MacCulloch; and this, besides being
accessible to only a few, having been published in 1830, made no mention of
colours, which form the most beautiful part of the art.
In the first year of the present century a treatise was written by a Captain
Jones, which has been copied, in whole or in part, into almost every work since
published. The greater portion of it is absurd and impracticable, and shows that
it was written by a person who undertook to teach what he had not learnt.
The first work of any real utility that came under my notice was a series of
papers by "Practicus." This was soon followed by another, varied by the new
chemical nomenclature.
The subject is far from being treated exhaustively in either of these works, so
that I trust the reader will find in the following pages a fund of information,
both in the repertory of recipes and the methods of manipulation.
To Chertier belongs the great improvement in colours. He was, as I was
informed by the late Mr. Southby, who knew him personally, and who derived
much information from him while in Paris, a retired French Artillery Officer, who
made colours "his study and theme." His "New Researches," published in 1856,
nearly thirty years after his first pamphlet, is an excellent work, that leaves little
to be desired in the way of colours. Gunpowder attracted but little of his
attention.
Tessier, of Paris, has written, since, his "Pyrotechnic Chemistry," and a new
edition of the "Firework-Maker," by Hutstein and Websky, was published three
or four years ago at Breslau.
I have neglected none of these sources of information, but do not know that I
have been able to learn much from them with which I was not previously
acquainted.
Gibbon's "Artillerist's Manual," and Benton's "Ordnance and Gunnery," published
at New York, have furnished me with a hint or two; also Owen's "Practice of
Modern Artillery," Scoffern's "Projectile Weapons of War," and his "New
Resources of Warfare;" but as all these works are on Military Pyrotechny, they
have not been available to any great extent. I mention their names, that
anyone desirous of a knowledge of Rockets and Shells, as instruments of
destruction, may know the books from which to gain the information.
It is possible that, as Chemistry advances, a few new substances may be
discovered; meanwhile, nearly every shade of colour may already be produced.
The extensive use of these colours, from their beauty and variety, for stars and
lancework, has very materially altered the class of fireworks, and necessitates
the employment of an enormous quantity of quick-match. The preparation of
this is one of the most disagreeable parts of Pyrotechny, besides demanding a
great amount of manipulative skill. Most amateurs are deterred from attempting
to manufacture it, and so have to content themselves with only the simplest
pieces.
I have, therefore, set myself sedulously to work to devise a different plan of
preparing it; and have succeeded—not till after many years, however, and
numerous failures, simple as it now seems—in discovering a process of
producing a splendid and perfect match, that leaves nothing to be desired;
which is easy and expeditious, and does not even soil the hands. Amateurs will
now find no trouble in making any pieces they may desire.
Dangerous chemicals, that might produce spontaneous combustion, have been
rigidly excluded; the fullest information throughout has been furnished for
preparing the mixtures with safety; the mode of projecting balloons, and the
manner of constructing steelyard scales, and every kind of tool used in the art,
has been added; and no pains have been spared to make the work as complete
and comprehensive as possible.
In conclusion, I may say that I have had thirty-five years' experience; and there
is nothing I have not tried repeatedly: the reader may, therefore, place the
greatest confidence in everything recommended. He has only to imitate to
succeed.
T. K.
 CONTENTS.
Page
Roman Candles:
To Make a 5⁄8 Roman Candle 1
To Make a Roman Candle Star 5
To Damp Stars 7
To Make Lac Solution 9
To Make Wax Solution 10
To Make Stearine Solution 10
To Make Gum Solution 11
To Make Dextrine Solution 11
To Make Paste 13
Roman Candle Scoops 15
To Charge Roman Candle Cases 19
To Make Touchpaper 23
To Make Slowmatch 24
To Make Quickmatch 24
A New Method 26
Rockets:
To Make a 6⁄8 Rocket 34
To Charge Rocket Cases 39
Rocket Stars 53
To Make Cut Stars 54
Dry Pill-box Stars 54
To Fill the Boxes with Dry Colour 56
Another Way 56
Bottomless Pill-boxes 57
Enveloped Stars 58
Rocket Heads 59
Wheel and Fixed Cases 66
Gerbes 69
Flower Pots 72
Port Fires and Shell Fuses 73
Tourbillions:
To Make a Tourbillion 75
Saxons 81
Five Pointed Stars 83
Squibs 84
Serpents 90
Pinwheels 91
Crackers:
To Make a Cracker 98
Leader Pipes 103
Maroons:
To Make a Maroon 105
Another Method 106
Gold and Silver Rains 107
Peacock's Plumes 108
Saucissons 109
Pearl Streamers 111
Blue Lights and Star Candles, or Star Lights 111
Prince of Wales' Feathers 112
Lances:
To Make a Lance 112
To Form a Device, or Design 115
To Preserve Steel Filings, or Cast-iron Borings 116
Shells:
To Make the Shells 119
Cylindrical Shells 125
Asteroid Rockets 131
Compound Fireworks: 135
To Fire a Girandole of 100 Rockets at once 153
Tools 154
To Construct a Steelyard Scale 155
Montgolfier Balloons:
To Cut the Gores 161
To Join the Halves Together 161
 To Paste Two Gores Together 161
To Paste the Pairs Together 162
To Make the Last Joining 162
To Wire the Balloon 163
Winged Rockets:
To Make Winged Rockets 168
To Construct a Slot-Tube, or Rocket-Guide 170
List of Prices 172
Concluding Remarks 173
Fuses:
Roman Candle 185
Rocket 185
Bursting Fire 185
Starting Fire 185
Wheel and Fixed Case 186
Squib and Serpent 186
Pinwheel 186
Saxon 187
Five-Pointed Star 187
Spur-Fire, for Flower-Pots and Star-Candles 187
Tourbillion 187
Bengal Light 188
Blue Light 188
Wasp Light 188
Portfire 188
Shell Fuse 188
Gerbe 188
Tailed, Streamer, or Comet Stars, for Rockets, Shells, and Roman
189
Candles
Oiled Tailed Stars for Rockets and Shells 189
Steel Stars for Rockets and Shells 190
Pearl Streamer 190
Gold Rain 190
Silver Rain 191
White or Bright Stars and Lances 191
 Sugar Blues for Stars and Lances 191
Blue Stars and Lances without Sugar 192
Crimson and Scarlet Stars and Lances 193
Green or Emerald Stars and Lances 194
Deep and Pale Yellow Stars and Lances 195
Mauve and Lilac Stars and Lances 196
Purple and Violet Stars and Lances 196
Magnesium Colours for Stars and Lances 197
Slow Fires, to be Heaped upon a Tile in the shape of a Cone, and Lit at
197
the Top
THE PYROTECHNIST'S TREASURY.
 ROMAN CANDLES.
To Make a 5⁄8 Roman Candle.
Procure a straight piece of brass tube, 6⁄8 of an inch external diameter, and
161⁄2 inches long. Saw or file off a piece, 11⁄2 inch long, fig. 1. This is for the
star former, and is drawn of the correct size.
In the other piece, of 15 inches, fix a handle, as shown, in diminutive, in fig. 7.
This is for the case former. It should be filed smooth at the end.

Take another piece of brass tube 41⁄2/8 or 9⁄16 of an inch external diameter, and
about 161⁄2 inches long. In this also fix a handle, or fix it into a handle, fig. 4.
Invert it, and set it upright in a flower-pot, filled with sand or loose mould. Melt
some lead in a ladle, and pour it slowly into the tube, leaving room for the air
to escape up the side, till it is full. If the lead is poured in rapidly, the confined
air, expanding, jerks the metal up, and may cause serious injury. A pound or
more of lead will be required. When cold, drive the end of the lead in with a
hammer, and file it smooth. This is for a rammer.

Take a piece of deal, fig. 6, about 12 inches long, 6 broad, and 3⁄4 thick; and,
on the top, screw a handle, like one on a copper-lid, or black-lead brush. This is
for a rolling-board. An iron door-handle would serve, and may be obtained at
any ironmonger's, for about 2d. or 3d. A wooden one, however, about an inch
thick, not cylindrical, but slightly flat, and rounded at the edges, is preferable,
as it gives more purchase for the hand.
Cut a piece of tin, or zinc, or thin board, into the shape of fig. 8, in which the
distance between the arms a and b, across the dotted line, shall be 7⁄8 of an
inch. This is for a gauge, with which to measure the external diameter of the
case. Write upon it, "7⁄8 space."
Procure some 60 lb., 70 lb., or 84 lb. imperial brown paper: the size of a sheet
will be 29 inches by 221⁄2. Cut a sheet into 4 equal parts, each 141⁄2 by 111⁄4:
paste the 4 pieces on one side, and lay them one on another, with the pasted
face upwards, putting the fourth piece with the pasted side downwards, upon
the pasted side of the third piece. Turn them over: take off the now top piece,
and lay it flat on the near edge of a table, pasted side upwards. Take the
former, fig. 7, and paste the tube all over. Lay it along the edge of the paper,
bend the paper over with the fingers of both hands, and roll it tightly up, until
the external diameter of the case about fits the gauge, fig. 8. If the paper
should be too long, of course a piece must be cut off; if it should not be long
enough, more must be added, taking care to bind in the second piece with 3 or
4 inches of the first piece; for if the whole of the first piece be rolled up before
beginning the second, the latter, when dry, will probably slip off, and spoil the
case. The case having been rolled up, take the handle of the former in the left
hand, lay the case flat on the near side of the table, take the rolling-board, fig.
6, in the right hand, press the front part of it on the case, and drive it forwards
5 or 6 times, like a jack-plane, letting the handle of the former slip round in the
left hand. This will tighten the case, and render it, when dry, as hard as a book-
cover.
The former must always be pasted, before rolling a case, to prevent its sticking.
It should, likewise, be wiped clean with a damp sponge, before being laid aside.
Brass tubes keep clean a much longer time if lacquered. To lacquer them, clean
them with very fine glass-paper; make them hot by the fire, till you can just
bear them on the back of the hand; then, with a camel's-hair pencil, wash them
over with thin lac solution. The cases may be either 141⁄2 or 111⁄4 inches long;
but 111⁄4 is the best, for when the cases are too long, the fuse, as it
approaches the bottom, is apt, if slow, to smoke; if fierce, to set the top of the
case in a flame. If the learner decides upon 111⁄4 inches, the former and
rammer may each be 2 or 3 inches shorter.
After the first case has been rolled up to fit the gauge, it may be unrolled, and
the paper measured. Future pieces of the same quire of paper can then be cut
of the right size at once, so that the case will fit the gauge without further
trouble.
A large slab of slate is convenient for rolling upon; but a smoothly planed board
will answer every purpose.
When a number of cases are finished, hitch a piece of flax two or three times
round each of them, and hang them up to dry, in a place free from draught,
that they may not warp.
Flax is sold in balls; the thick yellow, at 2d., is the best. It is named,
indifferently, flax, or hemp. It is much used by shoe-makers; and is sold at the
grindery, or leather shops. Two or three thicknesses of this, waxed, or drawn
through the hand with a little paste, is very convenient for passing round the
necks of small choked cases, tying cases on wheels, &c.
 To Make a Roman Candle Star.

Take the former, fig. 1, which, as said before, is 11⁄2 inch long; have a
cylindrical piece of turned wood, box, beech, or mahogany, fig. 2, about 2
inches long, and of a diameter to just fit easily into fig. 1. At a point a, at the
distance of about 7⁄8 of an inch from the end d, with a bradawl, or very small
gimlet, or nosebit, make a hole, and drive in a piece of brass wire, to project
just so much as to prevent the tube slipping over it. A piece of a brass rivet,
such as used by shoe-makers, is convenient for the purpose. The part with the
head on is best; a quarter of an inch length will be sufficient, filed or cut off
with the nippers. It is evident that upon inserting fig. 2 into the tube fig. 1, a
vacant space of 5⁄8 of an inch will be left at the bottom. Fig. 3 is a piece of
turned wood, or, better still, of turned brass, exactly like fig. 2 without the side-
pin a. Now to pump a star, insert fig. 2 in fig. 1; press the tube into damped
composition, turn it round, and withdraw it. Rest the tube on a flat surface,
insert fig. 3, and give it two or three taps with a small mallet, like fig. 26. A
convenient size for the mallet is 11⁄2 inch square, 3 inches long, with a turned
handle. The mallet is best made of beech or mahogany. The slight malleting
consolidates the star, and prevents it from getting broken in charging; it will
compress it to about 41⁄2/8 or 9⁄16 of an inch in height. Push it out and set it by
to dry.
Stars are best made in summer, and dried in the sunshine; when dry they
should be put into clean pickle-bottles, furnished with tight-fitting bungs. A
piece of wash-leather passed over the bottom of the bung, gathered up round
the sides, and tied at the top like a choke, makes a good stopper. Shot, shaken
up in bottles, with water, soon cleans them.

To Damp Stars.
Stars containing nitrate of strontian must be damped, either with lac solution,
or wax solution; anything containing water destroys the colour. Nitre stars may
be damped with gum water, dextrine solution, or thin starch. Most other stars
with either of the solutions. Crimsons and greens will mix with boiled linseed oil,
but they cannot then be matched, as oil renders meal-powder almost
uninflammable. With all stars, not a drop more of the solution should be used
than is sufficient to make the composition bind; and it is advisable not to damp
more than half an ounce at a time; this is particularly the case in using the lac
solution, as it dries rapidly; and if a large quantity of composition is damped,
and gets dry, and has to be damped over and over again, it becomes clogged
with the shellac, and the colour is deteriorated. If it should get dry, and require
a second damping, it is best to use pure spirit only, the second time.
Before mixing compositions, every article should be as fine as wheaten flour,
and perfectly dry. Nitrate of strontian, if purchased in the lump, should be set
over the fire, in a pipkin; it will soon begin to boil in its water of crystallization;
it must be kept stirred with a piece of wood, till the water is evaporated, and a
fine dry powder left. A pound of crystals will yield about 11 ounces of dry
powder, which should be immediately bottled. Even then, if used in damp
weather, it is best dried again, and mixed with the other ingredients while
warm. This second drying may be in a 6-inch circular frying-pan.
Articles, separately, may be reduced to powder, with the pestle, in a mortar. See
that it is wiped clean every time, as there is danger of ignition with chlorates
and sulphurets. When the articles are to be mixed, they may be put into the
mortar, and stirred together with a small sash-tool. A 3⁄8 inch is a convenient
size. The mixture must then be put into a sieve, and shaken in the usual way;
or it may be brushed through with the sash-tool. Return it to the sieve, and
brush or shake through again. As it lies in a heap, level or smooth it with the
blade of a table knife, or any straight-edge; if thoroughly mixed it will present a
uniform colour; if it appears darker in one part than in another, it must be sifted
again. A sieve with a top and receiver is very desirable, as nearly all mixtures
are either black or poisonous; the dust from star mixtures is very injurious to
the lungs. If a top and receiver cannot be readily purchased, both may easily be
constructed out of a sheet of millboard, fastened with a bradawl and waxed
yellow flax, and neatly covered with paper.
Mixtures may be damped on a Dutch-tile, a marble slab, or a slate without a
frame. They may be stirred about with a dessert knife, pressed flat, and
chopped, or minced as it were, and again pressed flat.

To Make Lac Solution.

Put 1⁄2 an ounce of flake shellac into a tin pot, and pour upon it a quarter of a
pint, or 5 ounces of methylated spirit; or, preferably, a like quantity of wood
naphtha. Let it stand for about a day, stirring it occasionally till dissolved. Then
half fill a basin with boiling water; set the tin containing the lac, in it, and leave
it till it boils and curdles. If the water does not remain hot long enough to make
it boil, set it in a second basin of boiling water. As soon as it has curdled,
remove it; and when cold, pour it into a vial, and cork it. Spirit must never be
boiled over a fire, nor near one, as the vapour might inflame. Keep the pot,
therefore, while in the hot water, at a distance from a fire, or flame of a lamp or
candle.

To Make Wax Solution.

Put into a vial 1⁄2 an ounce of white wax, (bleached bees' wax), pour upon it 5
ounces of mineral naphtha, (coal or gas tar naphtha), keep it tightly corked.

To Make Stearine Solution.

Dissolve a piece of composite candle in mineral naphtha, in the same way.
Mineral naphtha must not be used near a candle or fire, as it gives off an
inflammable vapour, at less than 100° Fahrenheit.

To Make Gum Solution.

There is no better way of preparing this than simply to put cold water upon
gum arabic, and let it stand till dissolved. If for sticking purposes, as much
water as will just cover the gum will be sufficient; but, for making quickmatch,
1 ounce or 11⁄4 ounce of gum to a pint of water. If required in a hurry, put the
gum into cold water, in a pipkin, or tin saucepan, set it on the fire, make it boil,
and keep stirring till dissolved. When cold, bottle, and cork it.

To Make Dextrine Solution.

Take half an ounce of dextrine, and 5 ounces, or a quarter of a pint of cold
water, put the dextrine into a cup or basin, add a little of the water, and mix it
well with a teaspoon, rubbing it till all is dissolved; then add the remainder of
the water, stir well together a second time, pour it into a vial, and cork for use.
Dextrine, wetted to the consistency of honey, may be used instead of thick
gum-arabic water, for pasting. For this purpose it is advisable to keep either in a
wide-mouthed bottle, and to set the bottle in a gallipot containing a little water;
the brush, a camel's-hair pencil, or very small sash-tool with 1⁄3 of the bristles
cut away on each side, to render it flat, can then be kept in the water, when not
in use; this will prevent it, on the one hand, from becoming dry and hard; and,
on the other, from getting clogged and swollen. It can be squeezed between
the thumb and fingers, when wanted for use. The flat gum brushes now sold,
bound with tin, are not pleasant to use, as the tin oxidises, and turns of a
disagreeable brown colour. If there is a difficulty in obtaining a graduated water
measure, one sufficiently correct for pyrotechnic purposes may be made with a
vial. Paste a narrow strip of paper up the outside of the vial, weigh 4 ounces of
water in a cup, in the scales: pour it into the vial, mark the height, and divide it
into 4 equal parts, for ounces; of course, it can be graduated into half and
quarter ounces, and increased, if large enough, to 5 or more ounces. A gallon
of distilled water weighs exactly 10 pounds. Consequently, a pint of pure water
weighs a pound and a quarter. This is also near enough for spirit, though, of
course, spirit is a trifle lighter. Doctors' vials are often marked with ounce
divisions.

To Make Paste.
Paste is most economically made in a zinc pot, which may be 4 inches deep,
and 31⁄2 inches diameter. Any zinc worker will make one to order for about 6d.
Put into it 2 ounces of wheaten flour, add a little cold water, rub the two
together with a spoon till smooth and free from lumps; pour in more water till
the pot is full within about an inch; set the pot in half a saucepanful of water,
put it on the fire; make the water boil, and keep it and the paste boiling for 4 or
5 minutes, stirring the paste the while. Remove it from the fire, and set it by to
cool. The paste is to remain in the zinc pot, in which it will keep good for a
length of time, and beautifully white.
Some recommend alum in paste, I think it best avoided, especially in cases
intended to receive coloured fires. Alum is a double salt, a sulphate of alumina
and potassa; it has an acid reaction; and, coming in contact with chlorate of
potash and sulphur, may cause spontaneous combustion. A drop of sulphuric
acid instantly ignites stars containing them. At theatres, the clown sometimes
fires a cannon, with what appears to be a red hot poker; but which, in reality, is
only a piece of wood, painted red. A mixture is made of chlorate of potash and
sulphur, or sugar, a glass bead is filled with sulphuric acid, and the hole stopped
up with wax. This is laid in the mixture, and when it is struck with the poker,
the liquid escapes, and inflames the potash and sulphur. Sulphate of copper is a
particularly dangerous salt, and must never be used, as it is almost certain to
cause spontaneous combustion. Chertier, to whom pyrotechny otherwise owes
so much, introduced an empirical preparation, by dissolving sulphate of copper
in water, together with chlorate of potash, drying it, and wetting it with
ammonia: but this, however dried, when again wetted, turns litmus paper red.
Practicus has named it Chertier's copper. I discommend its use.
Two paste brushes will be sufficient for an amateur, sash-tools, one about an
inch diameter, the other smaller for light purposes. Let them stand in the paste.
If they get dry, the bristles fall out. For convenience, one may be kept in the
paste, and one in water.
Dry clay, powdered and sifted as fine as possible, is used for plugging, or
stopping up the bottoms of cases. I have, for some time, discontinued its use,
and employ plaster of paris in preference. Directions will be given for each, so
that the learner can adopt which he pleases: but plaster is infinitely preferable.
It is an American improvement.

Roman Candle Scoops.

No species of fireworks require greater care in their construction than roman
candles. In the first place the stars must be fierce, that they may light
thoroughly: next, they must not be driven out with too great velocity. For this
purpose the blowing-powder must be carefully adjusted. The stars, also, must
be of so easy a fit that when put into the case they may fall to the proper depth
of their own accord. If they require pushing, they are too tight, and will
probably be blown out blind. When made as directed they will, necessarily, be
of an easy fit, as they will be of the inner diameter of the brass tube, while the
bore of the case is equal to its external diameter.
To regulate the blowing powder, prepare a number of little scoops, like fig. 5,
which is about the right size for the bottom star. They are formed of pieces of
tin, zinc, or copper. Cut a long strip of tin, 1⁄2 an inch broad; cut this across into
7 pieces, of the following lengths, 15⁄8, 16⁄8, 17⁄8, 2, 21⁄8, 23⁄8 and 4 inches.
Round off the corners. Take a piece of brass wire, or stair-rod, about 1⁄4 inch
diameter, and with the wooden mallet, before mentioned, fig. 26, bend each of
the pieces round the rod into a half cylinder, or gutter. Take up the smallest,
and hold 5⁄8 of an inch of the end of the stair-rod in the end of the semicylinder,
to keep it open; put the other part from a to b, fig. 5, in a vice, and pinch it up;
it will assume the form represented; the bowl part will be 5⁄8 long, and the
handle 1 inch long. Make the bowl of the next scoop 6⁄8 of an inch long, the
next 7⁄8, and so on; the handle will always be 1 inch long. The last, for the top
star, will have a bowl of 3 inches. The smallest scoop ought to hold as much
grain powder as will weigh about 1⁄12 of the star; but to have the scoops
accurate, it will be necessary to charge a roman candle, fire it, and observe
whether the stars go a uniform height. For measuring the interval fuse, or fuse
between the top of one star and the bottom of the next, a large scoop of the
size of fig. 10 will be required. The tin may be an inch broad, and the bowl part
21⁄2 inches long, bent round the rammer, fig. 4. To adjust it, take a roman
candle case, fit it on the foot, fig. 9, which is a piece of wood, or brass, turned
with a tenon to fit tight in the bottom of the case. Fill the scoop, and strike it
level, with a straight-edge; empty it into the case, rest the foot on a flat
surface; insert the rammer, fig. 4, and jolt it up and down, a dozen times, or
more, lifting it about 1⁄2 an inch at a time; put in another scoopful, and jolt it in
like manner. If the 2 scoopfuls, thus compressed, fill an inch of the case, the
scoop will be correct. If more or less, the scoop must be shortened, or
lengthened, accordingly.
A piece of writing-paper may be pasted and wound twice round the handle of
each scoop, as from a to b, fig. 4. One dot can be put upon the scoop, for the
first, or bottom, star; two dots, for the second scoop, &c., or any memorandum
can be written upon them, for future guidance. Should they get soiled, they
may be cleaned with a soaped damp piece of sponge.
Gunpowder, for fireworks, is used in two forms; meal-powder, and grain-
powder. Meal-powder is a fine black dust, and is employed in all cases of
mixing. Grain-powder is of three kinds, F, FF, and FFF, fine, double fine, and
treble fine. FFF is best for crackers, simply because it runs rapidly down the
pipes: for driving stars, shells, &c., F will be sufficient; but FFF may be
employed: FF need not be purchased. All kinds of powder may be obtained of
Pinnell, 214, Whitechapel Road. If, in any place, there should be a difficulty in
obtaining meal-powder, F grain-powder may be crushed in a leather bag, by
laying the bag on a hard surface, and beating it with a hammer. The leather
should be of the same kind as shoes are made of.

To Charge Roman Candle Cases.

Pour some F grain-powder into a wooden bowl, or platter, represented by fig.
11. Round the edge lay the little blowing-powder scoops, side by side,
beginning with the smallest at a, the next at b, and so on to g. Put some roman
candle fuse into a large tin scoop, made to stand on a flat bottom, like fig. 12,
the same, in shape, as used by tea-dealers; and, on the right-hand of it, lay the
charging-fuse scoop, fig. 10. If the roman candle is to contain different-coloured
stars, set seven in a row, in the order desired. When the cases are intended to
be fired in threes or fours, the stars in one may be all blue, in another crimson,
in another green, in another white. Fit the foot, fig. 9, in the bottom of the
case, put in a scoopful of clay, insert the rammer, fig. 4, and jolt it till the clay is
well compressed. The clay should fill 1⁄2 an inch. This being done, invert it, and
shake out any little dust that may remain. Put in the little scoopful a, of F grain-
powder; then lay the scoop at a. Now put in a star. As previously stated it ought
to fall of its own accord; but make sure that it has reached the blowing-powder,
by putting in the rammer. Having ascertained this, put in a scoop of fuse, fig.
10; lay the scoop on the left of fig. 12; insert the rammer and jolt it; put in
another scoop of fuse, fig. 10; lay the scoop on the right of fig. 12; insert the
rammer and jolt it, as before. Then proceed with the scoop b of grain-powder,
and lay it at b, and so on, till the case is filled. The fuse on the top star is best
driven in with a short solid rammer and mallet, as it is difficult to jolt the long
rammer in so small a space. The last 1⁄8 of an inch, near the mouth of the case,
should be fine meal-powder, as it binds better than the roman-candle fuse, and
also blows off the leader pipe.
The blowing-powder scoops, having been laid at a, b, &c., all that is required is
to turn the bowl or platter, a little round to the left, and they will come in
rotation, ready for the next case. Also, by putting the scoop, fig. 10, alternately
to the left and right of the scoop, fig. 12, it will always be known whether the
proper quantity of fuse has been put in.
Coloured stars, from their fierceness, have a tendency to burn in the cases. This
defect may be remedied by putting upon each star a small scoopful of Starting
Fire, No. 1, before putting in the interval fuse; as much as will fill round the
sides of the star. This composition is somewhat fiercer than would suit for the
regular fuse, so catches the blowing-powder sooner.
A roman candle is well charged when the stars isochronise, or come out at
equal intervals of time: they should, also, theoretically, ascend to equal heights;
but, with coloured stars, this cannot be perfectly insured, as some shrink more
than others in drying, and, of course, fit more loosely; some are heavier, some
fiercer than others.
The interval fuse must always be driven in at twice, never at once. Each star,
with its blowing-powder and fuse, occupies about an inch and a half; perhaps a
trifle more.
Instead of driving in clay at the bottom, plaster of paris may be used, and then
the foot, fig. 9, will not be required. Have some plaster of paris in a wide-
mouthed bottle; a glass of cold water with a salt-spoon in it; and a number of
pieces of paper, about 4 inches square. Put a small quantity of the plaster on
one of the pieces of paper; indent the middle with the finger; put to it a little
water, and work it up with a dessert-knife. Just as it gets to the consistency of
mortar, and is about to set, mould it with the fingers, to the shape of a cork;
push it into the end of the case; rest the case on a flat surface; insert the
rammer, and give it two or three slight jolts; turn it round a few times, and
withdraw it. If the plaster sticks to the end of the rammer, it shows, either that
you have used the plaster too wet, or have not turned the rammer round a
sufficient number of times.
No more plaster must be mixed at a time than will suffice for one case. When
plaster has once set, it cannot be mixed up a second time; therefore take a
fresh piece of paper, and let the knife be cleaned every time. It is advisable to
have two dessert-knives, then one can be used with which to scrape the other.
As much plaster should be used as will fill the case up about 1⁄2 an inch. They
must be set by to dry; their not requiring the use of the foot will be found a
great convenience.
Roman candles are usually made from 3⁄8 to 6⁄8; but 5⁄8 is a very satisfactory
size. If a roman candle is intended to be fired singly, twist a piece of touchpaper
round the mouth. If the cases are intended to be fired in threes, fours, &c., to
form a bouquet, or to be placed round a mine, jack-in-the-box, or devil-among-
the-tailors, omit the touchpaper, and envelope the case in double crown, made
to project an inch beyond the mouth, to receive the leader, or quickmatch.
A steel-pen inserted, nib backwards, in the end of a small paper tube, rolled
round the end of a penholder, makes a neat little scoop. It may be fastened in
with a little plaster of paris. A scoop may also be made with a quill.
Cracks in wooden bowls may be stopped with the same material; and, if
painted over with linseed oil, after getting dry, will remain waterproof for a long
time. A screw may be made to hold in brickwork, by drilling a hole in the brick,
and pushing in the screw, covered with plaster.

To Make Touchpaper.

Dissolve 1⁄2 an ounce of nitre in 1⁄2 a pint of hot water. Procure some 12 lb.
double crown blue; cut each sheet into four equal parts, 15 by 10. Lay them
smooth upon each other, and, with a sash-tool dipped into the nitre solution,
wash them over on one side, and hang them up to dry.

To Make Slowmatch.

Dissolve 1 dram of nitrate of lead in 1⁄2 an ounce of boiling water. Cut a sheet of
blotting paper into six equal parts, and wet them on both sides, with a sash-
tool, with the solution. When dry, paste a piece all over, and upon it smoothly
press another piece; upon this, pasted, put a third piece; and so on, till all the
six form a stiff board. Lay them under a heavy weight; and, when dry, with a
sharp knife and straight-edge, cut the whole into strips 1⁄4 of an inch broad.
Four inches will burn about a quarter of an hour. Narrow tape, boiled in the
solution, makes excellent slowmatch.

To Make Quickmatch.
Put into a pan 1 lb. of grain-powder, or meal-powder; pour upon it some thin
hot starch, and stir it well about, breaking all lumps, till the mixture is of the
consistency of paint. Procure some lamp-cotton, such as forms the wicks of
candles. It will probably consist of sixteen or twenty-four strands. Divide it
carefully into lengths of eight strands. This is not so easy a task as might
appear. The best way is to act the reverse of a man spinning string. Divide the
end of the cotton, say of sixteen strands, into two of eight each; fasten them to
two screw-hooks, a few inches apart. Take one in each hand, and walk
backwards, gently pulling them apart, and when they catch, untwist them: with
care they will separate without breaking or entangling. Drop the end of one of
the pieces into the pan; and, as it keeps falling, coil it round and round in the
mixture, and press it down with an iron spoon, until as much is pressed in as
the quantity can saturate. Be very particular that it is thoroughly soaked. Have
ready a wooden frame, fig. 13, of deal, resembling a swing looking-glass, with
the glass taken out. It may be 5 feet by 4. The frame is to be supported on
pivots between two uprights. Fix a nail or hook at the left-hand corner of the
frame, and tie the end of the cotton to it which has been hanging outside the
pan. Get a person to slowly turn the frame and hold it steady. Take hold of the
cotton in the right hand, shut the hand, and allow the cotton to slip gradually
and slowly through it, as the frame is turned; squeeze it very gently, so as to
allow it to come out well coated, and contrive to make it as round as possible.
When all the cotton is wound upon the frame, spread some sheets of paper—
old newspapers—on the floor; at each corner place a brick; lay the four corners
of the frame upon the four bricks; sift dry meal-powder all over the match, turn
it over and sift over the other side. Prop the frame against the wall, and leave it
to dry. One ounce of white starch will be sufficient for a pint and a half of water.
Rub the starch up smooth with a little of the water, then add the rest, and boil
it.

A New Method.
Put into a gallipot, or basin, some hot starch, made as before directed, or some
cold gum-water, or cold dextrine solution; and with a small stencil-brush, or a
3
⁄8 sash-tool, cut across the middle to make the bristles short and stubby, stir in
some meal-powder, till it is well mixed and looks like black paint. To avoid
repetition, it will be convenient to refer to this under the name of Meal-paste.
Take two towel-horses, fig. 14, and set them parallel 5, 10, or 20 feet apart. In
the top rail of each, drive four nails 1⁄2 an inch asunder. Fix the dry cotton to
the nail a, carry it across to the opposite nail b, pass it round the second nail c,
bring it to the opposite second nail, and so on, till the cotton lies in four parallel
lines, like the strings of a harp placed horizontally. Hitch the cotton, without
cutting it, to the hook of the weight, fig. 15. This weight is made by taking a
piece of brass tube, 1 inch diameter, and 4 inches long. Stop one end with a
bung, fill it with melted lead; and before it sets push in a lucifer match, having
previously cut off the priming. As soon as the lead is cold, pull out the match,
which having been partially burnt smaller, will come out easily, and in the hole
left by it screw a cup-hook, as drawn. Invert the weight, pick out the bung, and
fill its place with more melted lead. Now procure two pieces of planed deal
board; one 4 inches square, the other 6 inches square. With the left hand hold
the smaller piece close underneath the cotton; and with the sash-tool or stencil-
brush, work the meal-paste well into the fibres of the threads, pressing the
cotton on the board till it is thoroughly soaked, and rolling it over, laterally, to
make it as round and smooth as possible. It is best to begin on the left, and
work towards the right. If the wetting slackens the threads, pull them tight.
Now brush some meal-powder through a fine sieve, to free it from lumps; put a
tablespoonful or two upon the larger board; hold it close under the four
threads, as the other, with the left hand, and move it laterally forwards and
backwards, and down the whole length, at the same time brushing the meal
over the threads, with a soft, dry, sash-tool, till they are smothered, and giving
them an occasional jar to shake off the superfluous meal. By a little practice
they may be made as smooth and as round as a piece of wire. Leave the weight
hanging to them till they are dry. Instead of four nails on each rail, a dozen may
be put; and if the towel-horses are set 12 yards apart, 144 yards may be soon
made. In this case, three or four heavier weights would be required. These
contrivances, however, are by no means necessary for an amateur; four hooks
or nails opposite four others, anyhow supported, will be sufficient; and six or
eight feet apart is a good distance. A small quantity may thus be made one day,
and a small quantity another; and for this purpose it is best to use gum-water,
as it is always ready, and a little can be added to the dry left the day before,
and a little fresh meal stirred in. Three or four threads of white darning-cotton,
which is of two twists, make very good match; knitting or crochet cotton, which
is of three twists, produces, alone, excellent match; two or three pieces of the
first, or two of the latter, put into a leader pipe, side by side, blow through with
a violent report. Match may also be made of the very narrowest white tape, 3⁄16
of an inch broad; this, from its flatness, is peculiarly suited for enveloped stars.
For general purposes, however, lamp cotton is decidedly the best, as it is most
loosely twisted, and therefore the most absorbent. If match is liable to be
exposed to damp, it should be cased in thick leaders, and be prepared with
starch; in other respects, nothing can surpass gum-water. Of course iron pound-
weights, with a ring in them, will answer the purpose, or an iron pestle tied to
the cotton, or anything heavy; but the kind I have advised are most convenient.
They will weigh about a pound and a quarter each. The weight may hang over
a chair-back.
Match, to be perfect, ought, when cut across, to look black throughout; it
should, also, be stiff, straight, and round; but, to test it, cut off about 10 inches;
put one end of this into a leader pipe, so that 5 inches will be in the pipe, and 5
out. Hold the end of the leader with a pair of tongs, or lay it on the ground;
light the naked end. If the match is good it will burn gradually, though swiftly,
till it gets to the pipe; it will then blow through with a bang. The nearer the
leader pipe fits the match the better, only it ought to be large enough to allow it
to go easily in, without force: if laid in a roman candle case it would hardly puff:
the smaller and smaller the tube becomes the louder and louder the report: the
increase of power being in the inverse ratio of the diminution of space. A train
of gunpowder laid in the open air, or confined in a tube, comports itself in the
same way, as is well known. Match, when dry, should be kept straight; for, if it
gets broken, it acts like a cracker, snapping at every break.
A piece of zinc rainwater pipe, of suitable length, furnished with a bottom and
lid, or a couple of bungs, is convenient for keeping it in; but a far better
contrivance is a deal box, 5 feet long, 3 inches deep, and 3 broad, made of 1⁄2-
inch pine. The lid of this box can be readily furnished with three or five hinges,
like a piano, only made with string instead of brass. Take two pieces of string,
fig. 57, and tie them in a knot near the end, as at k. At x, fig. 58, make a
bradawl hole through the back of the box, near the top, and push the strings
through it; make another hole through the lid, put the string through it, and tie
in a knot on the top, as at z: and so with the other four. This will render the
explanation easier to understand; but, practically, a single piece of thicker string
is best, if you can manage to tie the second knot in the exact spot you wish.
The knots must not be too close to each other, as the lid requires a little play.
For safety, the box ought to be furnished with a lock.
 ROCKETS.
Rockets are charged in choked cases, on a spindle, to leave a hollow up the
middle, through which the fire may be communicated to nearly the whole of the
composition at once: this causes the sudden generation of an enormous
quantity of elastic vapour, which, being unable to escape instantly through the
contracted aperture left for its exit, exerts its pressure in the contrary direction,
and hurries the rocket forward. A stick, attached to it, guides it, like the rudder
of a vessel, or tail of a bird, or fish; while its weight and leverage keep the
centre of gravity a little below the case, and prevent the rocket from pitching
over. Its manufacture, therefore, from the commencement of cutting the paper
for the case, to its finish of fitting on the stick, requires an accurate adjustment
of all its parts. Disregarding the trade names of pound, ounce, &c., which, now
that moulds are dispensed with, are useful only for enabling the makers, from
tradition, to understand each other; the internal diameter, instead of the
external, as formerly, may be selected, from which to compute the relative
measures. Taking the bore of the case as unity, the proportions will be as
follow:—

1 Internal diameter of case.

11⁄2 External diameter.
8 Length of case.
6 Length of spindle.
3
⁄5 Bottom diameter of spindle.
64 Length of stick.
1
⁄2 by 1⁄2 Thickness and breadth of stick.

These proportions are most readily calculated by taking the diameter in eighths
of an inch. Selecting, for instance, a 6⁄8 rocket, we have, 6⁄8, inner diameter;
half as much again, 9⁄8, outer diameter.

Taking the numerator as inches: 6 inches, length of case; 3⁄4 of this, or, which is
the same, 1⁄2 the outer diameter, as inches, 41⁄2 inches, length of spindle;
putting 20 for the denominator, instead of 8 (8 being 2⁄5 of 20), 6⁄20 of an inch,
bottom diameter of spindle; 6 × 8, (numerator multiplied by denominator) = 48
inches, length of stick; 3⁄8 by 3⁄8, size of stick.
The following table exhibits the usual sizes in inches.

Bottom
Name Length Length Length Size
Inner Outer Diam.
of of of of of
Diameter. Diameter. of
Rocket. Case. Spindle. Stick. Stick.
Spindle.

10 5/8
1
10/8 10/8 15/8 10 7 ⁄2 10/20 times by
8=80 5/8

9 41⁄2/8
9/8 9/8 131⁄2/8 9 63⁄4 9/20 times by
8=72 41⁄2/8

8 4/8
8/8 8/8 12/8 8 6 8/20 times by
8=64 4/8

7 31⁄2/8
7/8 7/8 101⁄2/8 7 51⁄4 7/20 times by
8=56 31⁄2/8

6 3/8
1
6/8 6/8 9/8 6 4 ⁄2 6/20 times by
8=48 3/8

5 21⁄2/8
5/8 5/8 71⁄2/8 5 33⁄4 5/20 times by
8=40 21⁄2/8

41⁄2 21⁄4/8
41⁄2/8 41⁄2/8 63⁄4/8 41⁄2 33⁄8 41⁄2/20 times by
8=36 21⁄4/8

4 2/8
4/8 4/8 6/8 4 3 4/20 times by
8=32 2/8

3/8 3/8 41⁄2/8 3 21⁄4 3/20 3 11⁄2/8

times by
8=24
 11⁄2/8

To Make a 6⁄8 Rocket.

Have the former, fig. 7, a brass tube 6⁄8 of an inch external diameter; the
gauge, fig. 8, with a 9⁄8 aperture; procure some imperial brown paper, 70 lb. or
84 lb., the thicker and heavier the better. The best kind is made of old ropes, is
air-dried, and rough. This is not easily obtained now. The smooth machine-
made, cylinder-dried answers very well. The sort used for laying under carpets,
and which is 5 feet broad, is almost equal to the original rough imperial, and
should be procured, if possible, in preference to the smooth. Cut it into strips 6
inches wide. Paste the pieces well, and roll the cases as hard as possible, with
the rolling board fig. 6, till they fit the gauge. Lay them by for a few hours to
get partially dry. They must, then, be choked or strangled, about half a
diameter from one end of the case, that end that lay nearest to the left hand in
rolling, till they assume the shape of the neck of a vial. For this purpose have a
cylindrical piece of deal, alder, or any kind of wood, fig. 24, about 9 or 10
inches long, made to fit easily into the case; cut it into 2 pieces, a and b; b may
be an inch and a half long; round off the ends, just cut. In a, fasten a screw,
the solid wire, or uncut part of which, is 1⁄4 of an inch thick; saw, or file off the
head, and slightly taper the part projecting; in the piece b bore a hole to just fit
the wire of a. Slip b upon a, and push them into the case, so that the interval
between the two reaches within about 1⁄2 an inch of one end of the case;
slightly draw out b, to leave a neck, or hollow, round which to form the choke;
fix a staple, or screw-eye in a post: tie to it one end of a piece of cord, about a
yard long, and 1⁄10 of an inch thick; fasten the other end round the middle of a
stick, for a handle; take hold of the stick with the right hand; hold the case in
the left; pass the cord round the space left vacant and gradually tighten it,
turning the case round and round, so as to pinch it in equally on all sides, till it
assumes the shape of fig. 23. Hitch a piece of string, about 1⁄16 of an inch thick,
a few times round the neck, until it is filled up flush with the other part of the
case, remove the choker, and hang the case up to dry. A hitch is made by
simply bending the string backwards into a loop, like fig. 22; passing it round
the choke; pulling it tight; looping it again, again passing it round the choke,
and so on. The first loop is called a half hitch, and will not hold of itself; the
second loop completes the hitch, and makes it hold: the remaining hitches are
for filling up the choke, and restoring the cylindrical shape. Practise, by hitching
the string upon your thumb, 4 or 5 times: it will hold after the second loop; but
push it off at the end, and it will all fall loose again.
A case must not be choked when wet, or it will tear; nor must it be too dry, or it
will be difficult to choke it at all. The drier, however, it can be choked, the
better. Experience is the only guide. If properly dry, the wrinkles of the choke
will be small, and perfectly regular. Should the string stick to the case, chalk it;
but this is not likely to occur unless the paper is too wet. Have a foot, fig. 16,
turned of ash, or beech, or box, with a hemispherical nipple, 6⁄8 of an inch
diameter, as drawn. Bore it with a twist drill, or nosebit, to the depth of an inch
and a half. Procure a brass, iron, or steel wire, but preferably brass, 6 inches
long, and 3⁄10 of an inch diameter, and perfectly straight. Cut, on one end of it,
a screw, 11⁄2 inch long, fix it in a vice, wet the screw with glue, and screw the
foot on. File the 41⁄2 inches gradually tapering. The object of tapering it is
simply to make it deliver. A cylindrical hollow up the rocket would answer as
well, but the spindle could not be got out. The more conically true it is tapered,
of course, the better. Finish it off with a very fine file, and smooth it with glass
paper. The block into which it is screwed may be larger than drawn; the bottom
should be turned slightly concave, to make it stand firm, on the same principle
as the bottoms of plates, cups, &c., are made with a rim. The block and spindle
are better if cast in gun-metal, in one solid piece: the pattern to give to the
caster should be in one solid piece of wood. After casting, the rough parts must
be filed smooth; not many turners will be found willing to undertake to turn it.
The next articles required are a setting-down piece, fig. 17; three hollow
rammers, or drifts, figs. 18, 19, and 20; and one solid rammer, fig. 21. They are
simply cylindrical pieces of wood, turned with a head, to bear the blows of the
mallet. Beech, or box free from knots, will answer. The lengths of figs. 18, 19,
20, and 21, are to be 8, 6, 4, and 2 diameters, respectively, exclusive of the
head; that is, 6, 41⁄2, 3, and 11⁄2 inches. The hollow drifts are to have a
cylindrical, not conical, hole, bored up them, with a nosebit, or twist drill, to
within one inch of the handle, so as to clear the spindle by 1⁄2 an inch, to allow
of any dust being driven up it. The hollow in figs. 17 and 18 must be large
enough to fit the spindle loosely at a; in fig. 19, a trifle smaller, to fit loosely at
b; and in fig. 20, a trifle smaller still, to fit loosely at c.
The next requisite is a mallet, fig. 25, which may be of ash, or beech. It may be
a cylinder, 5 inches long, and 3 inches diameter, with a handle about 5 inches
long, and 1 inch diameter. Let it be turned at the end slightly concave, like the
bottom of fig. 16, that it may be set to stand upright, like a wine bottle. Or the
head may be made 3 inches square, like fig. 26; or, a small carpenter's mallet,
about a pound and a quarter weight, will answer.

To Charge Rocket Cases.

The first requisite is a solid block of wood, 6 or 8 inches square, and 18 or 20
high. This is indispensable. A piece of an old oak gate post, answers well. It
must be set upon the ground, or on a flag stone, not on a floor. It is impossible
to ram a rocket properly on the floor, because of the vibration. It is, also,
necessary to sit, not stand, at the work.
Put the setting down piece, fig. 17, into the case; press the case over the
spindle, and give the head a few blows with the mallet; this will smooth out the
wrinkles of the choke, which is all that fig. 17 is used for. Now put in a very little
powdered clay, and mallet it with fig. 18; as much clay as will reach up 1⁄12 of
an inch will be sufficient; its object is to preserve the choke from burning, and
getting enlarged. One cannot be too precautious with rockets. Now put in a
scoop of rocket fuse, insert drift, fig. 18, and mallet the fuse in firm, with about
a dozen and a half blows, or till it offers a resistance to the hand. The blows
must be light and numerous, not slow and heavy, like driving a post into the
ground. 18 blows with a momentum of 3, will consolidate the fuse: 3 violent
blows, with a momentum of 18, would perhaps bend the case, or drive the dust
up into your eyes. The mallet need not be lifted above 4 or 5 inches at a time.
If the rocket is not rammed firm throughout, it will, upon lighting, explode.
As soon as the case is charged about 11⁄2 inch, make a pencil or ink mark,
round the drift, where it stands level with the top of the case, for future
guidance; then charge another 11⁄2 inch with the second rammer, fig. 19, and
mark it in like manner; proceed in the same way with fig. 20. It is obvious that
if fig. 19 were used too soon it would get split by the spindle being driven up it,
and the spindle would be bent or broken, hence the advisability of marking the
drifts to know when to lay aside one, and take the other. Just before you get to
the top of the spindle, put in the solid rammer to feel how high the spindle
reaches near the top of the case; hold it by the thumb and finger to keep the
distance, and mark it down the outside of the case, by indenting the case with
the edge of the drift. Exactly 11⁄3 diameter, that is, in this case, exactly 1 inch,
above this indentation make another mark: then as soon as you have covered
the spindle, till you can no longer see it, with the use of fig. 20, begin charging
with the solid drift, fig. 21, till the composition inside is level with the top mark.
This being done drive in a little dry clay, till the case is full. Remove the rocket
from the spindle, by giving it a turn or two round to the right, not backwards;
then bore a hole through the clay, till you can see the composition, with a 3⁄16
inch shell-bit. The shell-bit should be fixed in a handle, and kept for the
purpose. It is not advisable to use a stock and bit, unless the bit is shielded, as
it is apt to bore too deeply. The bit may be fixed in a handle, by boring a large
hole in the handle, and pouring in melted lead, or pressing in plaster of paris.
Instead of driving in dry clay on the top of the composition, a little plaster of
paris may be pressed in; this, when dry, will allow of a perfectly clean hole
being bored through it; whereas the clay is apt to crumble, and chip out. The
object of the clay, or plaster, is to prevent the composition, which, containing
much charcoal, does not bind well, from getting disturbed, and the solid part
diminished, which would cause the stars to be ignited while the rocket was
ascending, or the fuse, perhaps, to blow through at the beginning. A piece or
two of naked quickmatch is to be inserted in the hole through the clay or
plaster, and a long piece is to be pushed up the core, or hollow, of the rocket,
as far as it will go; it is, then, to be cut off flush with the mouth, and fastened
to the side with a little dab of wetted meal powder, pressed on it with the blade
of a knife. If the rocket is intended to be lit with a port-fire, take a circular piece
of touch-paper, about 2 inches diameter, slightly paste it all over, lay it in the
left hand, press the mouth of the rocket down upon it, and smooth the edges of
the touch-paper up round the case. Otherwise, smear the end of the case with
the sash-tool dipped into meal paste, and when dry, paste a bit of touch-paper
round it, and twist to a point, like a squib. The appearance of the rocket is
shown at fig. 36; the dotted line round the mouth shows the touch-paper.
In driving with the hollow rammers, it generally happens that a little of the fuse
gets driven up the hole; this, if allowed to accumulate, is very troublesome to
remove; it should, therefore, be knocked out every time, by holding the drift in
the left hand, and giving the head a rap or two with the mallet.
The whole of the composition ought to be put in in about 12 scoops: try 2 or 3
scoops till you get one of the right size, then write upon the handle what-sized
rocket it belongs to. These directions may appear minute, but they will save
much trouble if attended to.
As it is convenient to know, beforehand, about what quantity of composition will
be required for any particular rocket, the following formula will be useful:—

E3 / 9 = drams.
Where E denotes the size of the rocket, in eighths of an inch.
Required the quantity of fuse for a 6⁄8 rocket.
(6 × 6 × 6) / 9 = 4 × 6 = 24 drams = 11⁄2 ounce.

For a 3⁄8 rocket?

(3 × 3 × 3)/9 = 3 drams.
Welcome to our website – the ideal destination for book lovers and
knowledge seekers. With a mission to inspire endlessly, we offer a
vast collection of books, ranging from classic literary works to
specialized publications, self-development books, and children's
literature. Each book is a new journey of discovery, expanding
knowledge and enriching the soul of the reade

Our website is not just a platform for buying books, but a bridge
connecting readers to the timeless values of culture and wisdom. With
an elegant, user-friendly interface and an intelligent search system,
we are committed to providing a quick and convenient shopping
experience. Additionally, our special promotions and home delivery
services ensure that you save time and fully enjoy the joy of reading.

Let us accompany you on the journey of exploring knowledge and

personal growth!

ebookname.com