en
0.25
0.5
0.75
1.25
1.5
1.75
2
Image Segmentation using Dual Distribution Matching
Published on Oct 09, 20124133 Views
We propose an image segmentation method that divides an image into foreground and background regions when the approximate color distributions for these regions are given. Our approach was inspired by
Related categories
Chapter list
Image Segmentation using Dual Distribution Matching00:00
Image segmentation00:04
Main idea of our method00:16
Challenges01:27
Agenda02:05
Background & Related works02:08
Binary labeling & energy optimization02:10
Consistency measure: local & global03:02
Overview of local & global measures (1)03:40
Overview of local & global measures (2)04:08
Methods based on local measures04:12
Overview of local & global measures04:43
Concept of global measures04:46
Issues on global measures05:12
Motivation of this research05:34
Proposed Method06:07
Proposed dual matching method06:10
Proposed new energy function06:47
Estimation of weights07:42
Optimal weight: proportional to area size07:46
Estimation of area size ratios08:37
Optimization09:14
Previous optimization method09:16
Overview of optimization process09:47
Experiments10:33
Common setup10:35
Experiment 110:57
Evaluation of estimated weights10:59
Purpose of this experiment11:36
Evaluation of weight estimation (1)12:01
Evaluation of weight estimation (2)12:13
Experiment 212:33
Image segmentation12:35
Image segmentation result (1)12:59
Image segmentation result (2)13:18
Image segmentation result (3)13:30
Experiment 313:38
Purpose: global vs local measures13:41
Approximate input from block masks14:08
Comparison using block masks14:28
Dual matching vs single matching14:40
Estimated weights vs fixed weights14:49
Global measure vs local measure14:57
Conclusion15:15
Contributions15:18