Gabor wavelet based feature extraction for Chinese character recognition

By using the direction selection ability and frequency locality of Gabor filter, the corresponding relation between the width of stroke of Chinese characters and frequency of the filter is established. Considering the multi-direction of stroke and multi-peak values of stroke width, the two-dimensional Gabor wavelet is used to analyze the Chinese character image in multi-scale and multi-direction. Through drawing the output coefficients of filters as statistic features, the high performance of the character recognition is realized. The experiment results show that the kind of method is superior to other methods, especially on gray or binary character images recognition with low resolution.