Graph Cuts for Image Segmentation

Ph.D. Seminar

Meghshyam Prasad

Under Guidance of

Prof. Sharat Chandran


Abstract

In computer vision, segmentation is the process of partitioning digital image into multiple regions (sets of pixels), according to some homogeneity criterion. The problem of segmen- tation is a well-studied one in literature and there are a wide variety of approaches that are used. Graph cuts has emerged as a preferred method to solve a class of energy minimiza- tion problems such as Image Segmentation in computer vision. We will see comparison of few of these methods.

Downloads

Seminar Report

Seminar Presentation

Seminar Progress

Week Work done during week Discussion in Meeting Reading Attendance
22nd July --- 28th July Read Papers on Multi-label image segmentation and Non-frontal face recognition. Finalized topic i.e. Multi-label image segmentation. MaxFlow-Min Cut Theorem Present
19th Aug --- 25th Aug Understood Max flow problem in Optimization and its relevance in image segmentation. -- Energy Minimization in Segmentation --
26th Aug --- 1st Sept Read classical papers on Multi-label image segmentation. I got to know what is exptected in seminar from me. Markov Random Field(MRF) Present
2nd Sept --- 8th Sept Read tutorials on MRF and its application in image segmentation. Discussed different approaches of formulating image segmentation problem as Graph cut problem. Normalized Cuts Present
16th Sept --- 22nd Sept Read paper "Normalized cuts and Image segmentation". Discussed relationship between spectral graph theory and image segmentation. -- Present
30th Sept --- 6th Oct Preparation of slides on α-β swap algorithm -- -- --
7th Oct --- 13th Oct Detail study of α-β swap algorithm Problems in α-β swap algorithm α-expansion algorithm Present
14th Oct --- 20th Oct Preparation of slides on α-expansion algorithm Issues in α-expansion algorithm Optimality Properties of α-expansion algorithm Present
21st Oct --- 27th Oct Preparation of slides on Optimality Properties, Potts Model and Multiway-cut problem -- -- --
28st Oct --- 3rd Nov Study of Efficient algorithms for Graph Cuts Application of B-K algorithm on Image Segmentation problem Voronoi Based Preflow Push Algorithm Present
4th Nov --- 10th Nov Comparison of B-K algorithm and Push-Relabel algorithm TODO TODO TODO

References

Yuri Boykov, Olga Veksler, and Ramin Zabih: Fast Approximate Energy Minimization via Graph Cuts Download

Yuri Boykov and Vladimir Kolmogorov: An Experimental Comparison of Min-Cut/Max-Flow Algorithms for Energy Minimization in Vision Download

Umamahesh Srinivas: Markov Random Fields Download

J. Shi and J. Malik: Normalized Cuts and Image Segmentation Download

Chetan Arora, Subhashis Banerjee, Prem Kalra, and S.N. Maheshwari: An Efficient Graph Cut Algorithm for Computer Vision Problems Download