In this paper, a new method for the problem of shape representation and classification is proposed. In this method, we define a radius function on the contour of the shape which captures for each point of the boundary, attributes of its related internal part of the shape. We call these attributes as “depth” of the point. Depths of boundary points generate a descriptor sequence which represents the shape. Matching of sequences is performed using dynamic programming method and a distance measure is acquired. At last, different classes of shapes are classified using a hierarchical clustering method and the distance measure.
The proposed method can analyze features of each part of the shape locally which this leads to the ability of part analysis and insensitivity to local deformations such as articulation, occlusion and missing parts. We show high efficiency of the proposed method by evaluating it for shape matching and classification of standard shape datasets.
Shape Representation and Classification Using Boundary Radius Function
Asian Conference on Computer Vision ; 2007 ; Tokyo, Japan November 18, 2007 - November 22, 2007
2007-01-01
10 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Computer vision , shape matching , shape classification , boundary radius function Computer Science , Image Processing and Computer Vision , Computer Imaging, Vision, Pattern Recognition and Graphics , Pattern Recognition , Artificial Intelligence , Biometrics , Algorithm Analysis and Problem Complexity
Shape Representation and Classification Using the Poisson Equation
British Library Conference Proceedings | 2004
|British Library Conference Proceedings | 2000
|Affine-invariant curve normalization for object shape representation, classification, and retrieval
British Library Online Contents | 2001
|British Library Conference Proceedings | 2016
|