Handwritten numeral recognition method based on characteristic matrix similarity analysis

The invention discloses a handwritten numeral recognition method based on characteristic matrix similarity analysis. The method comprises the following steps: A, obtaining a to-be-recognized handwritten numeral image; B, converting the obtained original handwritten numeral image to a gray level image; C, converting the obtained gray level image to a binary image; D, performing smooth optimization, cutting, and zoom on the obtained binary image, to obtain a binary characteristic matrix of the to-be-recognized image; E, combining the obtained binary characteristic matrix with each standardized numerical characteristic matrix in pairs, to form characteristic matrix pairs, respectively performing similarity analysis; analyzing an obtained association coefficient result, the higher the association coefficient be, the higher the similarity be, and the standardized number corresponding to the characteristic matrix with the highest similarity being the number obtained by recognition. Through the method, a computer ca