Seal Identification Based on Pixel Distribution Probability with Three-Mode and Nine-Section

This paper studied a new method of seal identification. First, the image of seal is segmented into 9 equal-area quadrate sections respectively on three modes-gridding, horizontal bar and vertical bar. The pixel distribution probability of strokes in every section is calculated. It is the ratio of the pixels of strokes of every section to the pixels of seal and represents the structural and statistic feature of seal. Then, the geometrical mean distance product and the weighted correlation measure of pixel distribution probability between identified seal and standard seal can be got. According to above two indexes and the statistical principle, the result of seal identification can be finished. This method proved to be simple and effective by experiment. It can realize the accurate and rapid seal identification with computer.