Table of Contents
Partial Shape Similarity
0. Motivation
1. Edge Detection
2. Edge Grouping
3. Extraction of Contour Parts
4. Recognizing Shapes by Parts
5. Results
Structural Shape Similarity
0. Introduction
1. Shape Segmentation
2. Skeleton Computation
3. Skeleton Pruning
4. Recognizing Shapes by Structures
Contour Grouping Based on Local Symetry
0. Introduction
1. From Edges To Contours
2. Center points
3. Reference Shape Model
4. Particle Filter
5.1. Experimental Result 1
5.2. Experimental Result 2
5.3. Experimental Result 3
5.4. Experimental Result 4
5.5. Experimental Result 5
5.6. Experimental Result 6
5.7. Experimental Result 7
5.8. Experimental Result 8
6. Conclusion
7. Contour Grouping Video
Shape Segmentation Skeleton Calculation Skeleton Calculation

Skeleton Computation

Points along the skeleton are the centers of maximal disks.
Points (red) along the skeleton (green) are the centers of maximal disks (blue) tangent to the contour at two or more points (yellow).

A skeleton is composed of the centers of maximal disks, which are inscribed circles that are tangent to two or more points of the shape contour. Skeletons can be computed using any one of four major classes of skeleton growing algorithms: thinning algorithms, Voronoi diagram based algorithms, distance map based algorithms, and mathematical morphology based algorithms. Our method can be applied irrespective of which algorithm is used to create the skeleton.

One of the greatest advantages of our algorithm presented on the next page is that it can either be applied explicitly in a post-processing pruning step, or it can be applied implicitly during the tree-growing process itself. This quality of our algorithm offers much flexibility as to how much processing resource is to be used and when it is to be used, while still returning the same results every time. Though our algorithm can be applied implicitly, we will describe it here as a post-processing pruning step.