Off-line Handwritten Chinese Character Recognition Based On Dynamic Gross Periphery Directional Line Element Feature

Due to enormous numbers, complex structure of Chinese character and various styles of handwritten character, the recognition of handwritten Chinese character becomes a difficulty in research. At present, the recognition of off-line handwritten Chinese character is still in development stage, and the reliability and accuracy of the existing OCR(optical character recognition) technology are difficult to meet the actual demand. Therefore, it is particularly necessary to enhance the research of off-line handwritten Chinese character recognition.In this paper, we design a new method based on dynamic gross periphery directional line element feature for off-line handwritten Chinese character recognition. Two-stage classification strategy is employed in the recognition system for huge Chinese character set. Extract four-side stroke density feature from the character image and use RBF (Radial basis function) neural network as the classifier in coarse classification. The handwritten Chinese character image is segmented dynamically, and gross periphery directional line element feature and hypo-gross periphery directional line element feature are extracted. We use minimum distance classifier based on urban distance in fine recognition process. The experiments on different samples show that the method proposed in this paper is feasible.