Printed Chinese Character Recognition System And Achieve

Chinese character has thousands years in history, and it is used by most people in the world. However, Chinese character is not alphabetic, to input Chinese characters into computer quickly and efficiently is a key task in Chinese information processing. So, Chinese character auto-recognition is the best choice for the large numbers of existing documents. It has important practical value and theory meaning in Chinese information processing, OA, machine translation, AI and so on.Based on researching the current OCR systems and related technologies, the paper presents the Recognition of Printed Chinese Character system. The words are as following:Firstly, for single classifier not good at Chinese character classification, multi-classification with complementary features and different matching methods are designed which efficiently improve recognition rate.Secondly, for improving the quality of preprocessing and overcoming the shortcomings of global and local methods, this paper introduces a binarization algorithm based on character outline detection. In this method firstly gets the outline of the character image then fills the characters based on the information of original image and outline. Experiment results show that the proposed approach is faster than local method and robust to noise than global method.In addition, this paper has improved the key steps in the process of Chinese character recognition, and proposed some new approaches: 1) Based on component method, analyze the layout using mathematic morphologic;2) Propose an algorithm of integrated method of least square which is based on best character point in incline emendation of image;3) Proposed feature extraction methods for structure, connected body, closed area and stroke in Chinese character's feature extraction. Most of all, proposed a stroke extraction method based on direction weight for left-falling stroke and right- falling stroke in stroke feature extraction.In one word, it achieves a recognition rate of 95% and a speed of 6s for one hundred Chinese characters using a printed Chinese character recognition system based on above algorithms.