A parallel computation approach to two-dimensional shape recognition is proposed and illustrated. The approach uses parallel techniques for contour extraction, parallel computation of normalized contour-based feature strings independent of scale and orientation, and parallel string matching algorithms. The string matching can be applied in a manner independent of rotation. The implementation on the EREW PRAM architecture is discussed, but it can be adapted to other parallel architectures. Illustrated examples and experimental results are presented.
Bibliographical noteFunding Information:
Shapes are conventionally represented by feature vectors or primitives based on gray levels, and/or contours, and/or textures, and/or range data (in the case of 3D shapes). There is a basic difference in the considerations for feature selection in the case of sequential processing as compared with parallel processing. For sequential processing, the requirement for reasonable computation time motivates the preference of low dimensionality feature vectors. The ultimate is, probably, a unique number (1D feature vector) for each distinct shape. Parallel processing enables effective and efficient computation of high dimensional feature vectors. Parallel computable contour-based feature strings are proposed. The term feature strings is used, rather than feature vectors, because string matching techniques are being used in the classification process. The proposed approach to parallel algorithms for 2D shape recognition consists of two aspects. One aspect is the development of parallel computable contour-based feature strings for shape recognition. The second is the incorporation of parallel string matching techniques in the recognition scheme. Image contours (or boundaries) have been used in various aspects of computer vision. Some previously published contour-based features are described in Section 2.3. An algorithm for parallel contour extrac- * This work was supported in part by the New York State Science and Technology Foundation, Center for Advanced Technology in Telecommunications, Polytechnic University, Brooklyn, NY, in part by the Paul Ivanier Center for Robotics and Production Management, Ben-Gurion University, Beer-Sheva, Israel, and in part by NSF grant NSF-CCR-8908286. ~: Now with the Electrical and Computer Engineering Department, Ben-Gurion University, Beer-Sheva, Israel.
- Computer vision
- Parallel algorithms
- Shape recognition
- String matching
ASJC Scopus subject areas
- Signal Processing
- Computer Vision and Pattern Recognition
- Artificial Intelligence