Research And Application Of Chinese Calligraphic Character Recognition

Chinese calligraphy is the art of visual and hand writing, which carries thousands of years of Chinese civilization. With the development of digital technology, Chinese calligraphy is able to be preserved, shared and displayed in front of people in digital form. But due to the long-term historical change, many calligraphic characters changed a lot in their shapes, so it is difficult for common people to recognize them and a recognition tool is urgently needed. Traditional Optional Character Recognition (OCR) technology performs poor in calligraphic character recognition, so the research of Chinese calligraphic character image recognition has great value.Numerous collections of historical Chinese calligraphic works are digitized and stored in CADAL (China Academic Digital Associate Library), and a huge database CCD (Calligraphic Character Dictionary) is built, which contains character images labeled with semantic meaning. In this paper, the research of Chinese calligraphic character recognition algorithm is carried based on the CCD and the contributions are as follows:(1) The fast and accurate LSH based large scale calligraphic character image retrieval strategy is proposed. The strategy is used to execute the similarity searching of the recognized calligraphic character image. First, denoising, binarization and normalization is executed to the CCD’s image and the recognized image in the preprocessing period and the GIST feature is extracted. Then, using the LSH to perform the fast retrieval and get the top X similarities. Finally, the accurate retrieval is executed using the shape feature of the top X images and recognized image, and get the top N similarities.(2) The retrieval based Chinese calligraphic character image recognition is proposed. After the fast and accurate retrieval of CCD in (1), the recognition result is given based on the semantic probability which is computed according to the ranks of retrieved similar images.(3) Using the algorithm proposed in this paper, CADAL provides online Chinese calligraphic character recognition service. Also the algorithm is applied in the labeling system, which contributes to the building of CCD.