A Study On The Segmentation Of The Mixed Arranging Character

At present, most OCR systems are based on single character recognition. The characters segmentation is one of the key factors which affect character recognition in OCR system.. It affects the recognition rate directly.In this paper, here presented a method of feedback recognition for mixed arranging segmentation based on character language discrimination. At first, to segment the character introductorily by projection, so as to discriminate character language; then with the help of recognition technique to segment the characters further and precisely, to judge whether the character is punctuation or parts of the character, to merge the parts of characters, and the result is confirmed after the recognition. To extract the conglutination characters, judging its type so as to carry out different segmentation method according to conglutination Chinese characters or English characters, and then to transfer to the module of recognition to segment the conglutination characters. The results of the experiments show that the method has a satisfying segmentation effect for mixed arranging of Chinese and English characters.