Research And Implements Of Image Search Engine Based On Text

Along with social progress, the Internet technology has also been an unprecedented development. Internet development to today, has no longer simply to provide users with textinformation, the media also contains a large number of pictures, music and video, which makes us to get the contents becoming the rich and colorful. Meanwhile,with the continuous increase of the content of micro-blog, blog, space, circle of friends and other users, the demand for pictures showed a rapid growth. Greater demands were being placed on Image Search.Image Search also has some problems, althought the current search engine technology is developing rapidly.For example,inaccurate sort of search results, low query efficiency, no mature system of document scoring, no mature system of keyword,and how to improve the efficiency of indexing creation of the mass image, Researchers need to do in-depth research.According to the above problems, this paper is mainly based on the text of the image search engine for the research object,A design andimplementation of a more complete picture search engine is proposed.It needs to score for the documents in the search process, the paper makes a detailed description on the TF-IDF method, also it makes a deep research on the feature weighting, combining with the keyword weight, time factor, picture level to improve TF-IDF method, and verifies its feasibility. Use the distributed search engine server to ensure the load balance, improve the index and search efficiency, and make the system have a good scalability, easy to maintain.After designing and implementingthe Image Search Engine, we test on the efficiency ofindex,efficiency of search and accuracy of search results. The results show that the design and implementation of the image search engine is reasonable and feasible. At the same time, the research results of this paper have a certain reference value for the image search based on content and video search and audio search.