A Segmentation and Recognition System for Touching and Broken Numeral Strings Based on Viterbi Algorithms

Abstract

Currently, in many OCR applications, it is difficult to segment and recognize touching and broken characters. In this paper, a segmentation and recognition system based on Viterbi algorithms is proposed to solve such a problem for touching and broken machine-printed numeral strings. This system includes two steps of segmentation and recognition. In the second step, first, a segmentation method is adopted to find the character nonlinear segmentation paths by combining gray scale and binary information based on a Viterbi algorithm; then, a recognition method of using a Viterbi algorithm is adopted to dynamically combine and recognize the character candidates with their reliabilities generated from the recognizer. Some experiments on a financial document analysis and recognition system indicate that this Viterbi algorithms based method is efficient for segmentation and recognition of touching and broken numeral strings, and enhances the accuracy and robustness of the recognition system.