Web image retrieval model based on social network and latent structure analysis

• Main Authors: Bo-Hung Sung, 宋柏宏
• Other Authors: Jian-Hua Yeh
• Format: Others
• Language: zh-TW
• Published: 2009
• Online Access: http://ndltd.ncl.edu.tw/handle/13675718440046696560

碩士 === 真理大學 === 資訊工程學系碩士班 === 97 === Advances in Internet and Web-based computation have inspired the design of content based information retrieval system, which are getting important to the success of a multimedia environment. It is more and more common for user to encounter web-based image retrieval services while surfing on the Internet, so the retrieval for a large amount of images is a big challenge for the current research domain. In order to solve these questions, our goal of this research is to creating a new type of web-based content information retrieval system which will provide more precise query result of web images. This research will utilize latent topic discovery algorithm along with social network analysis theory to create a new indexing and search scheme of web images. Our research will focus on indexing and query framework for web images. There are two major issues in this paper, as listed below: (1)Focus on content-based information retrieval, the text or keyword-based retrieval model will not be discussed in our research. (2) The social network analysis theory introduced in our research is different from Google’s PageRank algorithm. We adopt network power centrality computation as the re-ranking processing of initial query result to achieve better performance.