Occluded shape (2-D) recognition using edge based features

This paper presents an algorithm for the recognition of 2-D objects in partially occluded environments. The proposed method uses the edge of an object to represent as well as recognize shapes using partial information. The algorithm is based on detecting significant edge parts exhibiting some distinct features of a shape contour and representing some portions of the contour of an object. A set of connected components are obtained from the query edge image, and each of them is represented in terms of a collection of edge segments. Chamfer matching is used to detect the similarity between a partial edge segment of the query and a portion of an object template. The performance of the proposed method is verified on a large number of query images containing multiple objects from the database with different levels of partial occlusion. These set of query images were artificially generated from the 2-D shapes provided in the MPEG7 database.