Chinese Seal Image's Extraction And Recognitionin Chinese Painting And Calligraphy

In recent years, in the wake of digital image processing and development of computer technical, content-based image retrieval (CBIR) gradually has become a very active domain of research. In this domain, more and more new practicable systems and research models are come out.In the domain of cultural heritage digitization, people accept the positive effect of computer technical used in the cultural heritage protect, develop and popularize, the quantity of digital cultural heritage startling increased, problem of information retrieval gradually appeared. But in this domain, application of CBIR is still limited, especially in the retrieval Chinese painting and calligraphy area. After analyzed, the main reasons are: Firstly, is because inheritable of cultural tradition. For Chinese painting and calligraphy, the motif, literary and painting skill, dyestuff, and use of paper, same kind of composition always has same low-level visual features. Secondly, since the subsistent of "semantic gap", lead to a different from low-level visual features to the requirement of retrieval in high-level semantics. And because of the plentiful cultural background, Chinese painting and calligraphy have contained more complicated semantic information.So it won't get good result if only use low-level visual features and CBIR to retrieval the image of Chinese painting and calligraphy.The proposition of disquisition is via extraction the key object's semantic information in the Chinese painting and calligraphy, to solve the problem of image information's retrieval. Firstly, after summarize great deal of basic knowledge and base skill on image processing and image analysis, the disquisition has extracted a method to use low-level visual features to extract Chinese seal in Chinese painting and calligraphy automatically. Secondly, extract a CBIR model for Chinese painting and calligraphy by using the semantic information in Chinese seals to retrieval. Finally, design and set up a Chinese seal image recognition system base on mature technical of CBIR. The system can get interrelated semantic information via recognition Chinese seal image, for example: character of seal, author, time, etc.Since the injection of Chinese seal's semantic information, it will decrease the affect of "semantic gap" to the system. The ability of retrieval base on CBIR will make a great progress.The main work of disquisition as follows:1. According to research on Chinese seal image, summarize six main characteristic of Chinese seal (character of seal), established base for design Chinese seal image extract method according to the characteristic.2. Summarized color space's relationship of transformation and low-level visual features, and also summarized formula in common use, algorithm and technique about measurement method of the similar degree of image. 3. Extracted a method for extraction Chinese seal in Chinese painting and calligraphy automatically. Set up a system for extraction Chinese seal in Chinese painting and calligraphy automatically.4. Used CBIR, design and develop a model system which used moment invariants to achieve Chinese seal recognition.5. Extracted a retrieval system model base on content-based Chinese painting and calligraphy, which used Chinese seal image semantic information to assist retrieval.6. The disquisition extracted the way and the plan to improve the method for extraction Chinese seal in Chinese painting and calligraphy automatically; and also to improve the model system which used moment invariants to achieve Chinese seal recognition.