EARLY FIRE DETECTION

SYSTEM USING IMAGE

PROCESSING
 Department of
Computer Systems Engineering (CSE)
MUST Mirpur Azad Kashmir

FINAL YEAR PROJECT PROPOSAL

GROUP MEMBERS

Ahsan Azam FA17-CSE-003 2017-2021

Marva Touheed FA17-CSE-007 2017-2021
Urooj Zubair FA17-CSE-014 2017-2021
Sharjeel Mukhtar FA17-CSE-057 2017-2021

SUPERVISOR NAME
Engr. Waqas Ahmed

CO SUPERVISOR NAME
Engr. Hafiz Atta-ul-Mustafa Ahtesham

Submission Date: 25 January 2021

 Abstract

Fire outbreak is a common issue in Malays and causes tremendous damages toward
nature and human interest. To reduce loss of life and property from fire, an early
warning is an imperative. Fire detection is the primary goal of this project other
than surveillance. The objective of the project is early detection of fire separated
from preventive measures to reduce the losses because of risky fire. The proposed
system is mainly based on image processing and arduino serial communication. In this
project, at the client end, the fire pictures will be feeded as video frames. The
proposed framework utilizes RGB, YCbCr and HSV color space. The benefit of
utilizing YCbCr color space is that it can isolate the luminance from the
chrominance more successfully than RGB color space. Along with this smoke,
motion, area detection is additionally performed utilizing its color attributes. . Fire
growth is analyzed and calculated based on frame differences. Flame plays a more
and more important role in the fire recognition. The proposed framework comprise
of equipment, for example, arduino, GSM to send message in case of fire detection.
There is a camera for the observation. This camera will give an ongoing video yield
to the client on the PC or PC by means of a little GUI. Thus the fire will be
detected utilizing this model. This project can likewise be served for security and
observation applications.

Problem statement
Fire hazards is the one of the most common hazards of every country. Fire,
particularly fire in buildings, can spread rapidly and cause great loss of life and
property. In this way, early fire detection and cautioning is necessary. Smoke
sensors and temperature sensors have been generally used to secure property and
give cautioning of flames. However, smoke and temperature sensors are slower
than light detectors, they take a lot of time to detect fire in case of large buildings.
These type of sensors cannot be used in open environment because the efficiency
can be effected by the external interference. To cover the large area, many smoke
or temperature sensors are required. Due to the rapid developments in digital
camera technology and video processing techniques, we purposed a system that
will help to detects the fire in early stages.
 Introduction

In this proposed system instead of analyzing characteristics features of fire i.e

color, area, motion, smoke individually, all the characteristics are analyzed
simultaneously to reduce the false alarm rates which was available in the past
detection systems. Initially, our system will extract images of environment on real
time basis. These pictures at that point go through the identification strategies of:
Area detection, Color detection, Motion detection and Smoke detection. The
proposed framework will give the combine result at the output if smoke or fire is
detected. The proposed system has been designed to distinguish fire in class A and
class C objects, as characterized in the National Fire Protection Association
Standard. Class A objects are characterized as being wood, cloth, Rubber, Plastic or
paper and Class C objects are electrical objects, for example, Power Boards and
Electrical Motors and so on. Class A and Class C have comparative fire qualities.
This system gets input as picture or video and produce output in the form of alarm
and sending a message to authorities in case of fire detection.
Background

The fire detection research based on video sequences can be classified into two
categories: fire flame detection and fire smoke detection [1]. Flame colors used to
detect fire are descibed in [2-4]. Wenhao and Hong [2] extracted flame objects by
iterative adaptive threshold techniques, and then used fire flame color as a part of
characteristic information analysis to detect fire. Juan et al. [3] proposed the
analysis and extraction of flame colors in the RGB-color space. Celik and Demirel
[4] used YCbCr colors space to separate luminance from chrominance. It is
possible for a false detection if only color characteristics have been used. There is
research regarding flame shape characteristics [1]. Junying and Xiaoxiao [5]
proposed a fire detection method by searching for sharp corners on the contour line
of multi-layer closed contours of flame images. Yan et al. [6] also developed flame
characteristic detection based on threshold and the dynamic contours of flames.
Since fire flames have so many shapes within a video sequence, it is likely to
receive less accurate visual information. There is extensive research using
comprehensive information on the characteristics of flame [1].

Methodology

As visual color pictures of fire have high values in the red component of the RGB
coordinates. This property permits simple threshold-based criteria on the red
component of the color images to segment fire images in natural scenarios.
However, not just fire gives high values in the red component. Another
characteristic of fire is the proportion between the red part and the blue and green
segments. An image is stacked into color detection system. Color detection system
applies the particular property of RGB pixels and gives the output result as a
picture with a selected area of color detection. For that, color space RGB and
YCbCr is picked. For grouping of a pixel to be fire we have defined seven rules. If
a pixel fulfills these seven rules, we state that pixel have a place to fire class. Color
alone isn't sufficient to distinguish fire. There are numerous things that share the
same tone as things that are not fire, for example, a desert, sun, red leaves and
different items. The key to recognize the fire and the fire shaded items is the nature
of their motion. Motion detection is used to detect any occurrence of movement in
a video. It is done by investigating distinction in pictures of video frames. There are
three principle parts in moving pixel detection: frame/background subtraction,
background registration, and moving pixel detection Similar to the fire detection.
We are likewise demonstrating smoke pixels. The smoke pixels do not show
chrominance attributes like fire pixels. Toward the start, when the temperature of
the smoke is low, it is normal that the smoke will show color from the range of
white- bluish to white. Close to the beginning of the fire, the smoke's temperature
increments and it gets shading from the scope of dark grayish to black. Utilizing
this thought, HSV color spaces is used. The output is noisy, thus the motion
property of the smoke can be utilized to eliminate such noisy parts. The proposed
system will activate alarm and send a message to the management in case whether
fire or smoke or both are detected.
The hardware required for the implementation of this project is listed below.

 Arduino Uno R3 (Just for the Communication)

 LCD 16*2
 I2c Module
 Camera
 Step Down Power Supply
 Bridge Rectifier
 Capacitors
 Diodes
 Relay
 Buzzer
 GSM Module
The software’s used in the implementation of this hardware are shown in the list
below.
Arduino Uno R3 (Just for the Communication)
 Arduino IDE
 Visual Studio
 Python
 Proteus
Conclusion

This project, Fire Detection System has been created utilizing Image Processing.
This system can apply image processing techniques to distinguish fire. This system
can be utilized to screen fire. The system works on real time, as it extracts frames
in every 2 seconds, it provides continuous monitoring. This system has high
efficiency as it has incorporated techniques of Area detection, Color detection,
Motion detection, and Smoke detection. For better performance outcomes use of
RGB, HSV and YCbCr color space is made in the detection techniques, as per their
suitability, efficiency and properties. In case of fire detection, the system will
activate alarm and inform through GSM.

Advantages
 The technique has many advantages such as early fire detection, high
accuracy, flexible system installation, and the capability to effectively
detect fires in large spaces and complex building structures. It processes
image data from a camera by algorithms to determine the presence of a fire
or fire risk in images. Therefore, the detection algorithm is the core of this
technology, directly determining the performance of the image fire detector.
 Cost of using this type of detection is cheaper.
 This type of system is greatly simpler.
 The response time of fire detection system is faster.
 A single camera can monitor large area.
 The fire source can be saved in the form of images or video.
 Fire confirmation without visiting the location.
 Future scope

By proper analysis, suitable location height and length for camera

installment can be decided, in order to remove blind-spot areas.

Project Schedule
Semester Time-line ( Months/) Tasks

Semester VII Jane 1.Data Collection

Feb 2.Hardware Designing
March 3. Testing
Semester VIII April 1.Verifications
May 2.Report Writing
June 3.Submisison of work

References

Xiong Hao, Cong Linhu, He Weixin “Video Flame Detection Using Convolutional
Neural Networks”, 2019 IEEE 4th International Conference on Image, Vision and
Computing.

Wang Yuanbin , Dang Langfei and Ren Jieying, “Forest fire image recognition
based on convolutional neural network”, Journal of Algorithms & Computational
Technology Volume 13: 1–11, 2019.

M. Trinath Basu*, Ragipati Karthik , J. Mahitha , V. Lokesh Reddy, “IoT based

forest fire detection system”,   International Journal of Engineering & Technology ·
March 2018.

Vijayalakshmi, S.R., Muruganand, S.: Smoke detection in video images using

background subtraction method for early fire alarm system. In: IEEE 2nd
international conference on communication and electronics system (ICCES), pp.
167–171 (2017)
Fu Tianju, Zheng Chang'e, Tian Ye. Forest Fire Recognition Based on Deep
Convolution Neural Network in Complex Background [J]. Computer and
Modernization, 2016 (3): 52-57.

Foggia P, Saggese A, Vento M. Real-Time Fire Detection for Video-Surveillance

Applications Using a Combination of Experts Based on Color, Shape, and
Motion[J]. IEEE Transactions on Circuits & Systems for Video Technology, 2015,
25(9):1545-1556.

https://www.youtube.com/watch?v=nh7xQa-qJCk

https://www.youtube.com/watch?v=nbIEz6kUgEg

https://www.youtube.com/watch?v=SjN_vCDi6_I

https://www.youtube.com/watch?v=rebFKSp-Ss0

https://www.youtube.com/watch?v=nh7xQa-qJCk&t=26s