Slide 1

1 P. Arbelaez, M. Maire, C. Fowlkes, J. Malik. Contour Detection and Hierarchical image Segmentation. IEEE Trans. on PAMI, 2011. Student: Hsin-Min Cheng Advisor: Sheng-Jyh Wang

Slide 2

Outline Introduction Contour Detection Hierarchical Segmentation Results Conclusion 2

Slide 3

Introduction Original ImageContour Contour 3

Slide 4

Introduction Original ImageSegmentation Segmentation 4

Slide 5

Introduction From Contour to Segmentation Original ImageSegmentationContour 5

Slide 6

Introduction Goal Contour Detection Hierarchical Segmentation from Contours Original ImageSegmentationContour 6

Slide 7

Outline Introduction Contour Detection Hierarchical Segmentation Results Conclusion 7

Slide 8

Contour Detection 1. Learn local boundary cues 2. Global framework to capture closure, continuity 3. Local Cues and global cues combination 8

Slide 9

Learn local boundary cues Image Local Boundary Cues Model Brightness Color Texture Cue Combination Contour Detection 9

Slide 10

Learn local boundary cues Brightness L*a*b* colorspace Color L*a*b* colorspace Texture Convolve with 17 filters Filters for creating textons 10 Contour Detection

Slide 11

11 Learn local boundary cues Oriented gradient of histograms Example Gradient magnitude G at location(x, y) Three scales of r 11 Contour Detection ure

Slide 12

12 Learn local boundary cues Local Cues Combination 12 Contour Detection ure

Slide 13

Global framework to capture closure, continuity Contour Detection 13 V:image pixels E:connections between pairs of nearby pixels =>Build a weighted graph G=(V,E) from image

Slide 14

Global framework to capture closure, continuity Contour Detection 14

Slide 15

Local Cues and global cues combination Contour Detection 15 Local CuesGlobal cues

Slide 16

Outline Introduction Contour Detection Hierarchical Segmentation Results Conclusion 16

Slide 17

Hierarchical Segmentation Multiple Segmentations Fixed resolution Hierarchy of Segmentations Flexible resolution adjustment 17

Slide 18

Hierarchical Segmentation 1. From contours to segmentation 2. Hierarchical segmentation by iterative merging 18

Slide 19

Hierarchical Segmentation From contours to segmentation Watershed Transform Concept 19

Slide 20

Hierarchical Segmentation From contours to segmentation Watershed Transform Example 20

Slide 21

Hierarchical Segmentation From contours to segmentation Watershed Transform 21 Boundary strength Artifacts Weight each arc

Slide 22

Hierarchical Segmentation From contours to segmentation Oriented Watershed Transform 22 WT OWT

Slide 23

Hierarchical Segmentation Hierarchical segmentation by iterative merging Hierarchical segmentation Example 23

Slide 24

Brief Summary 24 Original Image - Local cues - Global cues Oriented Gradient of histograms Contour Oriented Watershed Transform Iterative Merging Hierarchical Segmentation

Slide 25

Outline Introduction Contour Detection Hierarchical Segmentation Results Conclusion 25

Slide 26

Result 26

Slide 27

Result 27

Slide 28

Result 28 Evaluation of segmentation algorithmsEvaluation of contour detector BSDS300 Dataset

Slide 29

Outline Introduction Contour Detection Hierarchical Segmentation Results Conclusion 29

Slide 30

Conclusion A high performance contour detector, combining local and global image information A method to transform any contour detector signal into a hierarchy of regions while preserving contour quality 30

Slide 31

Reference P. Arbelaez, M. Maire, C. Fowlkes and J. Malik. Contour Detection and Hierarchical Image Segmentation. IEEE TPAMI, Vol. 33, No. 5, pp. 898-916, May 2011 P. Arbelaez, M. Maire, C. Fowlkes and J. Malik. From Contours to Regions: An Empirical Evaluation. In CVPR 2009. P. Arbelaez and L. Cohen. Constrained Image Segmentation from Hierarchical Boundaries. In CVPR 2008. 31

Slide 32

Outline Introduction Contour Detection Hierarchical Segmentation Evaluation Results 32

Slide 33

Boundary Benchmarks ODS : optimal dataset scale OIS : optimal image scale AP :average precision 33

Slide 34

Region benchmarks(1) Segment Covering Probabilistic Rand Index [Unnikrishnan et. al. 07] [Yang et. al. 08] Variation of Information [Meila 05] Distance Between two segmentations in terms of their average conditional entropy given by 34

Slide 35

Region benchmarks(2) CoveringRand Index Variation of Information 35

Slide 36

Additional Dataset 36