Research On Key Techniques Of Personalized Web Image Retrieval

At present, the development of information and multimedia technology enormously promoted the innovation and development of Web image retrieval technology. Meanwhile, multimedia information is becoming the most important information carrier on the network, the multimedia information mainly composes of digital images, audios and videos, and images take a dominant position in the network multimedia information. Therefore, how to rearch the interested information accurately becomes a hot project at present. Meanwhile, search engines don’t take the difference in the needs of users into account. Therefore, while the number of Web inmage increase rapidly, users have to use lots of time, and the image retrieval efficiency is very low. So, people need to be provided personalized service to get interested information quickly.According to the problems, this paper puts forward a research of individuation web image retrieval techniques. The traditional search engines don’t consider that the users interest are different, and they don’t provide individuation retrieval service, so the retrieval efficiency is not high. In order to solve the problem, a method for individuation web image retrieval based on user interest model is proposed. Firstly, the formalized definition of user interest model is provided. And considering the time could affect user’s interest information, this paper cites new fator, effectively combining the short-term interest and long-term interest. Then the user interest model combines the methods of explicit tracking and implicit tracking to improve user’s interest information and provide individuation web image retrieval.The user interest model can be successfully applied in web image retrieval, and greatly improve the efficiency of image retrieval.User’s interest transfer, which is how to tranfer the interest information between different users, is also a problem for individuation web image retrieval. At present, the technology is also called a personalized recommendation. According to this problem, an algorithm of individuation user’s interest recommendation is provided. This algorithm adopts SVD and K-mearns clustering, effectively avoid sparseity of the score, and resolve the problem of user’s interest transfer for individuation web image retrieval. And this could offer the serve of individuation user’s interest recommendation for new and old users.Finally, an individuation web image retrieval system is designed. The conclusions and the research prospects of the thesis are presented.