The study of handwritten words is tied to the development of recognition methods to be used in real-world applications involving handwritten words, such as handwritten texts, bank checks, and postal envelopes, among others. In this paper an approach for Farsi bank checks was proposed in which the legal amounts are used to aid the recognition results of courtesy amounts. The legal amounts which are set of some of 40 specified words are divided into their subwords. Some IF-THEN rules are extracted from validation set to validate the recognized digits. These rules are used to confirm, correct or reject the recognized digit. The experimental results reveal a recognition rate of 85.33% without the legal amounts, and a reliability rate 99.31% with a rejection rate of 3.67%.
展开▼
