Handwritten Chinese Character Fuzzy Recognition based on Pixel Distribution Probability with Segmentation Mode of Concentric Circle

This paper proposed a new method of handwritten Chinese character recognition. The character image is segmented into 10 sections by 10 equal-interval concentric circles. 4 segmentation modes can be formed with the combination of 10 different-radius circles. The pixel distribution probability of strokes of character in every section is calculated. The concentric circles segmentation is an ideal method since it is adaptive for the shift, zoom, incline and rotation of image. The annulus segmentation represents the minutely structural feature of character. The circle segmentation is fit for various handwritten Chinese characters. As double-optimized evaluation, the minimal distance product and the maximal fuzzy correlation measure based on the distance and the correlation coefficient of pixel distribution probability between recognized character and standard character can obviously increase the rate of character recognition.