Recognition of Chinese Characters Using Comprehensive Shape Features

Abstract

The reason that Chinese characters are more difficult to identify than alphabetic characters lies in the fact that the precedent has more characters and more flexible shape distortion in hand-written process.In this paper, a novel approach based on comprehensive shape features for the recognition of Chinese characters is proposed. In this approach,firstly, the comprehensive features including the contour,the skeleton ,the texture and the stroke order are acquired by an improved algorithm.Secondly, the features are composed as a vector with variational weights using supporting vector machine.Lastly,according to the standard Chinese characters list the input of hand-written character is recognized.