Xu Huaping, Wang Wei, Liu Xianghua Beihang University, China.

Slides:



Advertisements
Similar presentations
Bayesian Belief Propagation
Advertisements

A Modified EM Algorithm for Hand Gesture Segmentation in RGB-D Data 2014 IEEE International Conference on Fuzzy Systems (FUZZ-IEEE) July 6-11, 2014, Beijing,
Human Pose detection Abhinav Golas S. Arun Nair. Overview Problem Previous solutions Solution, details.
Smoothing 3D Meshes using Markov Random Fields
Contextual Classification by Melanie Ganz Lecture 6, Medical Image Analysis 2008.
Markov random field Institute of Electronics, NCTU
Chapter 8-3 Markov Random Fields 1. Topics 1. Introduction 1. Undirected Graphical Models 2. Terminology 2. Conditional Independence 3. Factorization.
1 A Markov Random Field Framework for Finding Shadows in a Single Colour Image Cheng Lu and Mark S. Drew School of Computing Science, Simon Fraser University,
ICME 2008 Huiying Liu, Shuqiang Jiang, Qingming Huang, Changsheng Xu.
J. Mike McHugh,Janusz Konrad, Venkatesh Saligrama and Pierre-Marc Jodoin Signal Processing Letters, IEEE Professor: Jar-Ferr Yang Presenter: Ming-Hua Tang.
Wei Zhu, Xiang Tian, Fan Zhou and Yaowu Chen IEEE TCE, 2010.
1 On the Statistical Analysis of Dirty Pictures Julian Besag.
Robust supervised image classifiers by spatial AdaBoost based on robust loss functions Ryuei Nishii and Shinto Eguchi Proc. Of SPIE Vol D-2.
1 Markov random field: A brief introduction Tzu-Cheng Jen Institute of Electronics, NCTU
A Wrapper-Based Approach to Image Segmentation and Classification Michael E. Farmer, Member, IEEE, and Anil K. Jain, Fellow, IEEE.
Abstract Extracting a matte by previous approaches require the input image to be pre-segmented into three regions (trimap). This pre-segmentation based.
Effective Gaussian mixture learning for video background subtraction Dar-Shyang Lee, Member, IEEE.
Jacinto C. Nascimento, Member, IEEE, and Jorge S. Marques
Real-Time Decentralized Articulated Motion Analysis and Object Tracking From Videos Wei Qu, Member, IEEE, and Dan Schonfeld, Senior Member, IEEE.
Rician Noise Removal in Diffusion Tensor MRI
Advanced Image Processing Image Relaxation – Restoration and Feature Extraction 02/02/10.
Image Analysis and Markov Random Fields (MRFs) Quanren Xiong.
SAR Imaging Radar System A fundamental problem in designing a SAR Image Formation System is finding an optimal estimator as an ideal.
 C. C. Hung, H. Ijaz, E. Jung, and B.-C. Kuo # School of Computing and Software Engineering Southern Polytechnic State University, Marietta, Georgia USA.
MULTITEMP 2005 – Biloxi, Mississippi, USA, May 16-18, 2005 Remote Sensing Laboratory Dept. of Information and Communication Technology University of Trento.
Competence Centre on Information Extraction and Image Understanding for Earth Observation Matteo Soccorsi (1) and Mihai Datcu (1,2) A Complex GMRF for.
RECPAD - 14ª Conferência Portuguesa de Reconhecimento de Padrões, Aveiro, 23 de Outubro de 2009 The data exhibit a severe type of signal-dependent noise,
Presenter : Kuang-Jui Hsu Date : 2011/5/23(Tues.).
Prakash Chockalingam Clemson University Non-Rigid Multi-Modal Object Tracking Using Gaussian Mixture Models Committee Members Dr Stan Birchfield (chair)
Compressive Sensing Based on Local Regional Data in Wireless Sensor Networks Hao Yang, Liusheng Huang, Hongli Xu, Wei Yang 2012 IEEE Wireless Communications.
1 Physical Fluctuomatics 5th and 6th Probabilistic information processing by Gaussian graphical model Kazuyuki Tanaka Graduate School of Information Sciences,
Tracking with Unreliable Node Sequences Ziguo Zhong, Ting Zhu, Dan Wang and Tian He Computer Science and Engineering, University of Minnesota Infocom 2009.
Automated Detection and Classification Models SAR Automatic Target Recognition Proposal J.Bell, Y. Petillot.
Urban Building Damage Detection From Very High Resolution Imagery By One-Class SVM and Shadow Information Peijun Li, Benqin Song and Haiqing Xu Peking.
Markov Random Fields Probabilistic Models for Images
1 Markov Random Fields with Efficient Approximations Yuri Boykov, Olga Veksler, Ramin Zabih Computer Science Department CORNELL UNIVERSITY.
Authors: Rupert Paget, John Homer, and David Crisp
IGARSS 2011, Vancouver, Canada HYPERSPECTRAL UNMIXING USING A NOVEL CONVERSION MODEL Fereidoun A. Mianji, Member, IEEE, Shuang Zhou, Member, IEEE, Ye Zhang,
The 18th Meeting on Image Recognition and Understanding 2015/7/29 Depth Image Enhancement Using Local Tangent Plane Approximations Kiyoshi MatsuoYoshimitsu.
Automated Detection and Classification Models SAR Automatic Target Recognition Proposal J.Bell, Y. Petillot.
Image Analysis, Random Fields and Dynamic MCMC By Marc Sobel.
1 Markov random field: A brief introduction (2) Tzu-Cheng Jen Institute of Electronics, NCTU
12/4/981 Automatic Target Recognition with Support Vector Machines Qun Zhao, Jose Principe Computational Neuro-Engineering Laboratory Department of Electrical.
MCMC (Part II) By Marc Sobel. Monte Carlo Exploration  Suppose we want to optimize a complicated distribution f(*). We assume ‘f’ is known up to a multiplicative.
SAR-ATR-MSTAR TARGET RECOGNITION FOR MULTI-ASPECT SAR IMAGES WITH FUSION STRATEGIES ASWIN KUMAR GUTTA.
A Dynamic Conditional Random Field Model for Object Segmentation in Image Sequences Duke University Machine Learning Group Presented by Qiuhua Liu March.
Automated Detection and Classification Models SAR Automatic Target Recognition Proposal J.Bell, Y. Petillot.
Motion Estimation using Markov Random Fields Hrvoje Bogunović Image Processing Group Faculty of Electrical Engineering and Computing University of Zagreb.
Efficient Belief Propagation for Image Restoration Qi Zhao Mar.22,2006.
Markov Random Fields (MRF) Spring 2009 Ben-Gurion University of the Negev.
Cell Segmentation in Microscopy Imagery Using a Bag of Local Bayesian Classifiers Zhaozheng Yin RI/CMU, Fall 2009.
Jianchao Yang, John Wright, Thomas Huang, Yi Ma CVPR 2008 Image Super-Resolution as Sparse Representation of Raw Image Patches.
Learning Hierarchical Features for Scene Labeling Cle’ment Farabet, Camille Couprie, Laurent Najman, and Yann LeCun by Dong Nie.
Edge Preserving Spatially Varying Mixtures for Image Segmentation Giorgos Sfikas, Christophoros Nikou, Nikolaos Galatsanos (CVPR 2008) Presented by Lihan.
Ch 6. Markov Random Fields 6.1 ~ 6.3 Adaptive Cooperative Systems, Martin Beckerman, Summarized by H.-W. Lim Biointelligence Laboratory, Seoul National.
Shadow Detection in Remotely Sensed Images Based on Self-Adaptive Feature Selection Jiahang Liu, Tao Fang, and Deren Li IEEE TRANSACTIONS ON GEOSCIENCE.
Zhaoxia Fu, Yan Han Measurement Volume 45, Issue 4, May 2012, Pages 650–655 Reporter: Jing-Siang, Chen.
6.8 Maximizer of the Posterior Marginals 6.9 Iterated Conditional Modes of the Posterior Distribution Jang, HaYoung.
Biointelligence Laboratory, Seoul National University
Bag-of-Visual-Words Based Feature Extraction
Markov Random Fields with Efficient Approximations
STEREO MATCHING USING POPULATION-BASED MCMC
Binarization of Low Quality Text Using a Markov Random Field Model
Markov Random Fields for Edge Classification
Shashi Shekhar Weili Wu Sanjay Chawla Ranga Raju Vatsavai
Outline Texture modeling - continued Julesz ensemble.
A Block Based MAP Segmentation for Image Compression
Random Neural Network Texture Model
Outline Texture modeling - continued Markov Random Field models
Presentation transcript:

Xu Huaping, Wang Wei, Liu Xianghua Beihang University, China

 Introduction  SAR Image Segmentation with a MRF model  Fast Segmentation of SAR Imagery by Fusing Optical Imagery  Conclusions

 Synthetic Aperture Radar (SAR) systems can acquire SAR imagery at all-climate, day and night. Image segmentation is an important technique of automatic interpretation of SAR imagery. However, the performance of SAR image segmentation would decline due to speckle noise.  Image segmentation based on the Markov Random Field (MRF) model attracts much attention for it adequately considers the tonal and textural characteristics of imagery. In spite of speckles, SAR image segmentation using the MRF model achieves a much better performance.

 Different images can provide additional information which may improve the performance of image processing, especially when images of different sensors are combined. Therefore, SAR and optical imagery combination can be explored to improve image segmentation performance.  In this paper, we investigate to alleviate the time-consuming problem to carry out SAR image segmentation with a MRF model using simulated annealing algorithm. Two strategies are proposed for fast segmentation: First, an optical image is applied to accelerate SAR image segmentation by selecting uncertain pixels which attend the SAR image segmentation. Second, a fast annealing strategy is proposed to the simulated annealing algorithm to shorten the time consumed in optimization.

 Suppose is the pixel intensities of a SAR image and its segmentation label field.  Image segmentation is to obtain the label field given the observed image. The maximum a posterior (MAP) probability method is adopted to achieve it:  The likelihood function and prior distribution need to be known. Energy function

 The observation model is the conditional distribution of the background clutter given the segmentation label field. Suppose all pixels obey independent identical distribution, then:  Rayleigh, Gamma and K distribution can be used to describe the observation model of SAR imagery. Observation Model

 The random field is Markov random field iff: (1) (2)  If is a two-dimensional MRF, the prior model can be expressed as follows: Prior Model Fig.1 second-order neighborhood System

 The flowchart of SAR image segmentation with the assistant of one optical image is illustrated in Fig.2.

 We artificially classify the optical image into three classes. Pick up a region of target pixels and a region of background pixels from the optical image, calculate their mean intensities: Optical Image Classification

 The simulated annealing algorithm with Gibbs sampler is employed to carry out image segmentation using a MRF model.  A label is defined as the predominant label if the label is shared by over half of its neighboring pixels:  Label update strategy A Fast Annealing Strategy DoesDoes

Fig.4 SAR image from TerraSAR-X Fig.5 Optical image from Quickbird Fig.6 Classification result of Fig.5 Fig.7 The result of Fig.4 Fig.8 The result of Fig.4 and Fig.5 Fig.9 The result of Fig.4 and Fig.5 with the fast annealing strategy

Segmentation results Pixel number of wrong segmentation Consuming time (s) Fig Fig Fig Evaluation of Results

 Two strategies are proposed to accelerate SAR image segmentation using a MRF model with the simulated annealing strategy.  The consuming time of image segmentation can be shortened by adding an optical image into segmentation or adopting the fast annealing strategy.  Better performance can be achieved by adding an optical image into image segmentation.