Optical character recognition system based on a novel fuzzy descriptive features

In today's business environment where the security of accepting checks can sometimes be a burden, many businesses are turning to verification services to provide security for funds accepted as checks. In this paper we propose a new recognition system for processing the optical code, the magnetic ink character recognition (MICR) code. located at the bottom of bank checks and in some security documents. The proposed method is based on the use of a novel fuzzy descriptive feature using cross correlation to classify the characters. This method is fast robust and is not affected by shift or distortion of characters after thresholding. A 100% recall rate for the trained patterns and a recognition rate of 100% for distorted and skewed patterns with less than 1/spl deg/ skewness were obtained.