High-Order Statistics and Its Application in Handwritten Character Recognition
REN Jun-ling1, GUO Jun2
1. Department of Computer Information System,Beijing Information Science & Technology,Beijing 100101,China;2. School of Information Engineering,Beijing University of Posts and Telecommunications,Beijing 100876,China
Abstract：Since the signals we have collected always don't meet the normal distribution,the signal processing method based on the high-order statistics becomes more popular.For the feature of the handwritten Chinese characters,its decentralization characteristic can not be described as the normal distribution.So a new method based on the high-order statistic is proposed to describe it.Three different calculational methods of high-order statistics are given in the paper,and the performance of the template matching based on these three methods are compared according to the experiments on HCL2004 handwritten Chinese characters database.At the same time,the validity of the high-order statistics based recognition method is approved.
任俊玲;郭军. 包含高阶统计量的手写汉字分类尺度[J]. 电子学报, 2005, 33(10): 1876-1878.
REN Jun-ling;GUO Jun. High-Order Statistics and Its Application in Handwritten Character Recognition. Chinese Journal of Electronics, 2005, 33(10): 1876-1878.
2005, Vol. 33 
