en
0.25
0.5
0.75
1.25
1.5
1.75
2
Convexity Shape Prior for Segmentation
Published on Oct 29, 20143059 Views
Convexity is known as an important cue in human vision. We propose shape convexity as a new high-order regularization constraint for binary image segmentation. In the context of discrete optimization,
Related categories
Chapter list
Convexity Shape Prior for Segmentation00:00
Binary Segmentation Energy00:02
Length Regularization Shortcomings01:02
Other Regularization Models01:24
Shape Convexity01:51
Related Work in Continouis Optimization - 102:20
Related Work in Continouis Optimization - 202:39
Related Work in Continouis Optimization - 302:47
Our Approach02:56
Our Segmentation Energy with Convexity Shape Prior03:08
Energy Formulation for Convexity Shape Prior - 103:23
Energy Formulation for Convexity Shape Prior - 203:52
Energy Formulation for Convexity Shape Prior - 303:53
Energy Formulation for Convexity Shape Prior - 403:54
Energy Formulation for Convexity Shape Prior - 503:54
Energy Formulation for Convexity Shape Prior - 603:55
Energy Formulation for Convexity Shape Prior - 704:14
Difficulties in optimizing04:59
Optimization05:56
Trust Region Overview - 106:25
Trust Region Overview - 206:42
Trust Region Overview - 306:44
Trust Region Sub-Problem07:07
Aproxximate TR sub-problem07:50
Trust Region & Dynamic Programming08:26
Evaluation of E convexity - 108:51
Evaluation of E convexity - 209:04
Evaluation of E convexity - 309:06
Evaluation of E convexity - 409:57
Experiments & Results - 110:07
Experiments & Results - 210:33
Experiments & Results - 311:01
Experiments & Results - 411:20
Experiments & Results - 512:01
Comparison with QPBO and TRWS12:19
Comparison with QPBO and TRWS for Compact Model13:16
Limitations of our method13:50
Conclusions14:29
Thank you!15:04