TITLE: HANDWRITTEN NUMERAL RECOGNITION BASED ON TEMPLATE MATCHING AND FEATURE EXTRACTION NAMES: Sung-Won Yoon (yoons1@stanford.edu) Seong-Taek Chung (bluesky@stanford.edu) GOAL: Design and implementation of handwritten numeral recognition algorithms. The idea and researches concerning the recognition of handwritten numerals have been around for quite some time. This idea caught our attention with the help of the suggestions in the class web page and the eigenimages and transformation techniques discussed in class. We propose to do a three step process of carrying out this idea. First, a well defined acquisition of handwritten numerals will be carried out. A collection of handwritten numerals will be carried out and put into a database of numeral images. These images will be foundation for the following template matching method. Second, template matching based on eigenimages and several other unitary transforms discussed in class will be implemented. These matching will be carried out on the subset of data already acquired and on newly acquired data as well. The third step would be the implementation of some single stage feature-based algorithms. This will enable the comparison of two different approaches to recognition - template matching and feature extractoin. Time permitting, robustness of the algorithms to noise, the effects of resolution changes of acquired images, and the rejection criteria will be further investigated. References: 1. Computer Recognition of Unconstrained Handwritten Numerals, Ching Y. Suen et al, Proceedings of the IEEE, Vol.80, No.7, pp.1162-1180, July, 1992 2. Recognition of Handwriteen Digits Using Template and Model Matching, Paul Gadner et al, Pattern Recognition, Vol.24, No.5, pp.421-431, 1991 3. Handwritten Numeral Recognition Based on Multiple Algorithms, F. Kimura and M. Shridhar, Pattern Recognition, Vol.24, No.10, pp969-983, 1991 4. Computer Recognition of Handwritten Numerals by Polygonal Approximations, T. Pavlidis and F. Ali, IEEE Transactions of Systems, Man, and Cybernetics, Vol.SMC-5, pp.610-614, November, 1975 P.S. Prof. Girod, your comments on this proposal would greatly help in the navigational direction of our initial step. Thank you.