Research And Application Of Supports Vector Machine For Chinese Character Image Recognition

Chinese Characters recognition is one aspect of pattern recognition field. SVM (Support Vector machine) , the theory of small-sample statistical learning proposed by Vapnik, is an important research direction in pattern recognition. Based on the idea of Experience Risk Minimization, it is a good learning method for especially studying small-sample prediction. A new structural learning method has been developed by SVM. It can well solve the construction issue of a high dimensional model of small-sample set. It can get a biggish theory meaning and practice value that the SVM theory is used for the off-line Handwritten Chinese Characters Recognizing.The primary contents of this paper are:(1) Chinese Characters Image pretreatment and feature extraction. The image gradation, binarization, filtering and size normalization of manuscript character images are introduced in this paper, the feature date is extracted by four methods. Characters statistic feature and strokes feature can be denotation by the four methods inosculated date. The inosculated date is more useful in SVM recognition.(2) Characters recognition machine. A recognition machine is designed based on the SVM method. The effect of recognition is satisfied by using small number of off-line hand written Chinese Characters samples.(3). SVM classifier kernel functions of manuscript character images and selection of parameters. On the basis of the analysis of kinds of kernel functions, compared the performance of different kernel functions through simulation and tried to select kernel functions for the recognition of manuscript character.The test and simulation indicate: The SVM method can be used practically in SCUT-IRAC off-line handwritten Chinese characters recognition. It is effective and scientific that the handwritten Chinese characters are recognized on the feature small-sample.