![]() | Only 14 pages are availabe for public view |
Abstract The basic functions of a pattern recognition system are to detect and extract common feature from the patterns describing the objects that belong to the same pattern class, and to recognize this pattern in any environmen and classify it as a member of one of the pattern classes under consideratio. There are two essential requirements for a real-time pattern recognition : firstly, a fast accurate set of algorithms for image analysis, and secondly, substancial computation power. The advant architecture development provides an opportunity to better integrate these two requirements. In this thesis, algorithms for 2-d shape analysis have been investigated, two sets of boundary descriotors were considered, the fourier descriptor, and the translation invariant transforms (TIT). These algorithms provide description of 2-d shapes in terms of features independent of any specific context, the hard-ware coplexity has been considered wih speed and accuracy such that the overal performance is optimized. The thesis has compared and enhanced these algorithms for the purpose of feature extraction significant improvements in comparison with the existing techniques in terms of the above balanced criteria between speed, accuracy, and hard-ware complexity. |