Chinese characters are the world's most widely used characters which have different characteristics from the letters; while the rapid development of information technology makes various forms of information. Therefore, to deal with character information effectively is a key issue in information processing. The current research of the majority of Chinese character recognition system is based on OCR system, and research results have made great achievements, however, the current Chinese character recognition system is still inadequate which requires a higher quality of the input images. In view of this, it is great significant for this article uses the advantages of local features invariant to make study on a new type of Chinese character recognition system.First, the paper analyzes the current Chinese character recognition system and sums up the difficulties and shortcomings of the system.Secondly, this paper presents a novel Chinese character recognition system based on image matching, which uses SIFT feature as Chinese character extraction. The whole system framework is designed to three main modules: the image feature extraction module, high-dimensional vector index module, similarity matching module, and then we do a further analysis and interpretation of each module and introduce the design algorithms of the main modules. Finally, according to the proposed system design, we implement the novel Chinese character recognition system. For the purpose of the research, this paper designs the tests to verify the feasibility of this system and analyze the characteristics of the system according to the experimental results. Lots of the experiments approve that Chinese character recognition system we proposed is feasible, SIFT features have the ability to represent Chinese characters and our system has a certain robustness of stretching, rotation, affine transform, complicated background, noises and so on. |