SlideShare a Scribd company logo
Company
LOGO
PROGRESSIVE TEXTURE
SYNTHESIS ON 3D SURFACES
By
Rupesh Shet, Eran Edirisinghe,
Helmut Bez
R.Shet@lboro.ac.uk
Agenda
1. Introduction1. Introduction
2. Texture Synthesis on surface (Greg Turk)2. Texture Synthesis on surface (Greg Turk)
3. Discrete Wavelet Transform (DWT) /
Embedded Zerotree Wavelet (EZW)
3. Discrete Wavelet Transform (DWT) /
Embedded Zerotree Wavelet (EZW)
4. Proposed Algorithms4. Proposed Algorithms
5. Experimental Result5. Experimental Result
Introduction
 The progressive texture synthesis on
3D surface is based on:
1) Texture Synthesis on Surface (Greg Turk)
2) Multi-resolution DWT decomposition of
sample texture.
3) Prioritising the DWT coefficient in
progressive texture transmission and
synthesis.
Application
Proposed Algorithm has application in
bandwidth and processing power
constrained application domain.
Such as
1) Remote Visualisation
2) Streaming
3) Distributive/collaborative games/animation
Texture Synthesis on Surface (Greg
Turk)
(a) 256 K point (b) User defined vector
1) User define sample points from low to high density is
created on surface.
2) Using repulsion method points are separated to each
other uniformly which later connected to from mesh
hierarchy (Shown in above fig (a)).
3) Subsequently user define vector is created on surface to
indicate orientation of texture pattern (Shown in above fig (b)).
4) Mesh vertices visiting are sorted based on vector field.
Texture Synthesis on Surface (Greg
Turk)
1) Each points is then scanned to determine the best
match colour using neighbourhood search in sample
texture.
2) Multi-level synthesis process produce higher quality
of texture which adapt coarse-to-fine refinements.
3) The colour used in neighbourhood matching are
taken from either one or two levels of mesh
hierarchy (Shown in above fig (a,b,c,d)).
(b) LEVEL 2 (c) LEVEL 1 (d) LEVEL 0(a) LEVEL 3
Results Of Greg Turk
Algorithm
This algorithm is limited in bandwidth adaptive
ransmission media in modern application visualisation.
DWT/EZW
LH3
HL3
HH3
HL1
LH1
HH1
HL2
LH2 HH2
LL3
EZW
1) Large wavelet coefficients are visually more important than smaller
wavelet coefficients.
2) In an embedded coding algorithm the encoder can terminate the encoding
at any point there by allowing a target bit rate or target distortion metric to
be met exactly.
3) On the other hand, given a bit stream, a decoder can cease decoding at any
point in the bit stream.
DWT
Proposed Algorithm
DWT/EZW
Location
Important Feature
Importance Feature of Proposed Algorithm:
1) We replaced Gaussian Pyramid by DWT pyramid.
2) Which is later used in EZW based coefficient prioritisation
(As shown as DWT and Priority Using EZW modules named).
3) This ensure seamless texture representation capability.
4) Texture encoding via EZW enables embedded texture
decoding capability and allows any intermediate texture
quality to be readily reconstructed.
5) Made available at the receiver depending on bandwidth
constraint.
Result 1
Progressive texture synthesized on the bunny by proposed algorithm using texture sample-1Progressive texture synthesized on the bunny by proposed algorithm using texture sample-1
Result 2
Progressive texture synthesized on the bunny by proposed algorithm using texture sample-2Progressive texture synthesized on the bunny by proposed algorithm using texture sample-2
Question?
DWT & EZW Based Prioritisation
Scheme
4 K points
16 K Points
64 K points
I (sample image)
LL1
LL2
LL1
LL3
EZW
Based prioritisation
Scheme
I (Sample Image) Write thing here
Back
Point Image Locations
X1
Y1
X2
Y2
Write thing here
Back

More Related Content

What's hot (20)

PPTX
Tutorial on convolutional neural networks
Hojin Yang
 
PDF
CNN
Ukjae Jeong
 
PPTX
Review on cs231 part-2
Jeong Choi
 
PPTX
Introduction to CNN
Shuai Zhang
 
PDF
Demystifying Neural Style Transfer
SEMINARGROOT
 
PPT
On Complex Enumeration for Multiuser MIMO Vector Precoding
TSC University of Mondragon
 
PDF
Learning Convolutional Neural Networks for Graphs
Mathias Niepert
 
PDF
PIMRC 2016 Presentation
Mohamed Seif
 
PDF
Pr045 deep lab_semantic_segmentation
Taeoh Kim
 
PDF
ujava.org Deep Learning with Convolutional Neural Network
신동 강
 
PDF
1D Convolutional Neural Networks for Time Series Modeling - Nathan Janos, Jef...
PyData
 
PPTX
convolutional neural network (CNN, or ConvNet)
RakeshSaran5
 
PDF
PR-284: End-to-End Object Detection with Transformers(DETR)
Jinwon Lee
 
PPTX
Convolutional neural networks deepa
deepa4466
 
PDF
04020470
bsbs1980
 
PDF
Neural Network as a function
Taisuke Oe
 
PPTX
DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...
Joonhyung Lee
 
PDF
Adaptive Channel Equalization using Multilayer Perceptron Neural Networks wit...
IOSRJVSP
 
PDF
AI&BigData Lab 2016. Александр Баев: Transfer learning - зачем, как и где.
GeeksLab Odessa
 
PDF
PR-108: MobileNetV2: Inverted Residuals and Linear Bottlenecks
Jinwon Lee
 
Tutorial on convolutional neural networks
Hojin Yang
 
Review on cs231 part-2
Jeong Choi
 
Introduction to CNN
Shuai Zhang
 
Demystifying Neural Style Transfer
SEMINARGROOT
 
On Complex Enumeration for Multiuser MIMO Vector Precoding
TSC University of Mondragon
 
Learning Convolutional Neural Networks for Graphs
Mathias Niepert
 
PIMRC 2016 Presentation
Mohamed Seif
 
Pr045 deep lab_semantic_segmentation
Taeoh Kim
 
ujava.org Deep Learning with Convolutional Neural Network
신동 강
 
1D Convolutional Neural Networks for Time Series Modeling - Nathan Janos, Jef...
PyData
 
convolutional neural network (CNN, or ConvNet)
RakeshSaran5
 
PR-284: End-to-End Object Detection with Transformers(DETR)
Jinwon Lee
 
Convolutional neural networks deepa
deepa4466
 
04020470
bsbs1980
 
Neural Network as a function
Taisuke Oe
 
DeepLab V3+: Encoder-Decoder with Atrous Separable Convolution for Semantic I...
Joonhyung Lee
 
Adaptive Channel Equalization using Multilayer Perceptron Neural Networks wit...
IOSRJVSP
 
AI&BigData Lab 2016. Александр Баев: Transfer learning - зачем, как и где.
GeeksLab Odessa
 
PR-108: MobileNetV2: Inverted Residuals and Linear Bottlenecks
Jinwon Lee
 

Viewers also liked (10)

PPTX
Record matching over query results from Web Databases
tusharjadhav2611
 
PPT
An adaptive algorithm for detection of duplicate records
Likan Patra
 
PPTX
novel and efficient approch for detection of duplicate pages in web crawling
Vipin Kp
 
PPTX
Linking data without common identifiers
Lars Marius Garshol
 
PDF
Duplicate detection
jonecx
 
PDF
A study and survey on various progressive duplicate detection mechanisms
eSAT Journals
 
PPTX
Deduplication
Lars Marius Garshol
 
PDF
Tutorial 4 (duplicate detection)
Kira
 
PDF
Progressive duplicate detection
ieeepondy
 
PDF
Efficient Duplicate Detection Over Massive Data Sets
Pradeeban Kathiravelu, Ph.D.
 
Record matching over query results from Web Databases
tusharjadhav2611
 
An adaptive algorithm for detection of duplicate records
Likan Patra
 
novel and efficient approch for detection of duplicate pages in web crawling
Vipin Kp
 
Linking data without common identifiers
Lars Marius Garshol
 
Duplicate detection
jonecx
 
A study and survey on various progressive duplicate detection mechanisms
eSAT Journals
 
Deduplication
Lars Marius Garshol
 
Tutorial 4 (duplicate detection)
Kira
 
Progressive duplicate detection
ieeepondy
 
Efficient Duplicate Detection Over Massive Data Sets
Pradeeban Kathiravelu, Ph.D.
 
Ad

Similar to Progressive Texture (20)

PDF
High Speed and Area Efficient 2D DWT Processor Based Image Compression
sipij
 
PPT
fault analysis using wavelet transform.ppt
gururaj009
 
PDF
D2Net Deep Denoising Network in Frequency Domain for Hyperspectral Image.pdf
OKOKPROJECTS
 
PDF
G0523444
IOSR Journals
 
PDF
Cb34474478
IJERA Editor
 
PDF
145 153
Editor IJARCET
 
PDF
Image compression using embedded zero tree wavelet
sipij
 
PDF
SECURED COLOR IMAGE WATERMARKING TECHNIQUE IN DWT-DCT DOMAIN
ijcseit
 
PDF
Architectural implementation of video compression
iaemedu
 
PDF
Digital watermarking with a new algorithm
eSAT Publishing House
 
PDF
Digital watermarking with a new algorithm
eSAT Journals
 
PDF
5G mmwaves - problems and solutions (graduation project)
Faculty of Engineering - Alexandria University
 
PDF
Dynamic Texture Coding using Modified Haar Wavelet with CUDA
IJERA Editor
 
PDF
40120140505005
IAEME Publication
 
PDF
40120140505005
IAEME Publication
 
PDF
40120140505005 2
IAEME Publication
 
PDF
A Wavelet Based Hybrid SVD Algorithm for Digital Image Watermarking
sipij
 
PDF
Iaetsd wavelet transform based latency optimized image compression for
Iaetsd Iaetsd
 
PDF
www.ijerd.com
IJERD Editor
 
PDF
Image Compression using Combined Approach of EZW and LZW
IJERA Editor
 
High Speed and Area Efficient 2D DWT Processor Based Image Compression
sipij
 
fault analysis using wavelet transform.ppt
gururaj009
 
D2Net Deep Denoising Network in Frequency Domain for Hyperspectral Image.pdf
OKOKPROJECTS
 
G0523444
IOSR Journals
 
Cb34474478
IJERA Editor
 
Image compression using embedded zero tree wavelet
sipij
 
SECURED COLOR IMAGE WATERMARKING TECHNIQUE IN DWT-DCT DOMAIN
ijcseit
 
Architectural implementation of video compression
iaemedu
 
Digital watermarking with a new algorithm
eSAT Publishing House
 
Digital watermarking with a new algorithm
eSAT Journals
 
5G mmwaves - problems and solutions (graduation project)
Faculty of Engineering - Alexandria University
 
Dynamic Texture Coding using Modified Haar Wavelet with CUDA
IJERA Editor
 
40120140505005
IAEME Publication
 
40120140505005
IAEME Publication
 
40120140505005 2
IAEME Publication
 
A Wavelet Based Hybrid SVD Algorithm for Digital Image Watermarking
sipij
 
Iaetsd wavelet transform based latency optimized image compression for
Iaetsd Iaetsd
 
www.ijerd.com
IJERD Editor
 
Image Compression using Combined Approach of EZW and LZW
IJERA Editor
 
Ad

Progressive Texture

  • 1. Company LOGO PROGRESSIVE TEXTURE SYNTHESIS ON 3D SURFACES By Rupesh Shet, Eran Edirisinghe, Helmut Bez [email protected]
  • 2. Agenda 1. Introduction1. Introduction 2. Texture Synthesis on surface (Greg Turk)2. Texture Synthesis on surface (Greg Turk) 3. Discrete Wavelet Transform (DWT) / Embedded Zerotree Wavelet (EZW) 3. Discrete Wavelet Transform (DWT) / Embedded Zerotree Wavelet (EZW) 4. Proposed Algorithms4. Proposed Algorithms 5. Experimental Result5. Experimental Result
  • 3. Introduction  The progressive texture synthesis on 3D surface is based on: 1) Texture Synthesis on Surface (Greg Turk) 2) Multi-resolution DWT decomposition of sample texture. 3) Prioritising the DWT coefficient in progressive texture transmission and synthesis.
  • 4. Application Proposed Algorithm has application in bandwidth and processing power constrained application domain. Such as 1) Remote Visualisation 2) Streaming 3) Distributive/collaborative games/animation
  • 5. Texture Synthesis on Surface (Greg Turk) (a) 256 K point (b) User defined vector 1) User define sample points from low to high density is created on surface. 2) Using repulsion method points are separated to each other uniformly which later connected to from mesh hierarchy (Shown in above fig (a)). 3) Subsequently user define vector is created on surface to indicate orientation of texture pattern (Shown in above fig (b)). 4) Mesh vertices visiting are sorted based on vector field.
  • 6. Texture Synthesis on Surface (Greg Turk) 1) Each points is then scanned to determine the best match colour using neighbourhood search in sample texture. 2) Multi-level synthesis process produce higher quality of texture which adapt coarse-to-fine refinements. 3) The colour used in neighbourhood matching are taken from either one or two levels of mesh hierarchy (Shown in above fig (a,b,c,d)). (b) LEVEL 2 (c) LEVEL 1 (d) LEVEL 0(a) LEVEL 3
  • 7. Results Of Greg Turk Algorithm This algorithm is limited in bandwidth adaptive ransmission media in modern application visualisation.
  • 8. DWT/EZW LH3 HL3 HH3 HL1 LH1 HH1 HL2 LH2 HH2 LL3 EZW 1) Large wavelet coefficients are visually more important than smaller wavelet coefficients. 2) In an embedded coding algorithm the encoder can terminate the encoding at any point there by allowing a target bit rate or target distortion metric to be met exactly. 3) On the other hand, given a bit stream, a decoder can cease decoding at any point in the bit stream. DWT
  • 10. Important Feature Importance Feature of Proposed Algorithm: 1) We replaced Gaussian Pyramid by DWT pyramid. 2) Which is later used in EZW based coefficient prioritisation (As shown as DWT and Priority Using EZW modules named). 3) This ensure seamless texture representation capability. 4) Texture encoding via EZW enables embedded texture decoding capability and allows any intermediate texture quality to be readily reconstructed. 5) Made available at the receiver depending on bandwidth constraint.
  • 11. Result 1 Progressive texture synthesized on the bunny by proposed algorithm using texture sample-1Progressive texture synthesized on the bunny by proposed algorithm using texture sample-1
  • 12. Result 2 Progressive texture synthesized on the bunny by proposed algorithm using texture sample-2Progressive texture synthesized on the bunny by proposed algorithm using texture sample-2
  • 14. DWT & EZW Based Prioritisation Scheme 4 K points 16 K Points 64 K points I (sample image) LL1 LL2 LL1 LL3 EZW Based prioritisation Scheme I (Sample Image) Write thing here Back