A fuzzy logic-based new method for recognizing handwritten persian numbers

Recognizing handwritten numbers can have many uses such as sorting mailers according to their P. O. Box number, automatic office books, sorting bank drafts, and so on. The present study is concerned with proposing a new method based on a blend of fuzzy logic, genetic algorithm, and simulated annealing, to process handwritten data sets. To recognize patterns of numbers with various sizes, lengths, and shapes, pre-processing will be an inevitably necessary phase. As a result, two stages of pre-processing, namely correctingslant characters and thinning them were used. The framing method was used to extract the specifications, whilegenetic algorithm was utilized to select effective features. The proposed method was tested on MODARES database, as a resource approved by researchers, while 96. 67of the recognized cases were found to be correct.