Based On The Design Of The Shape Of The Icons Of The Search Engine

This paper mainly focus on the development of shape feature searching ,including ordinary image search engine development, shape feature extraction ,as well as indexing system designing for search engine, also analysis the idea of shape feature symbol search engine development. Symbol search engine ,part of content-based image search engine , is a Combination of digital image processing and web 2.0.It involved with network application development ,digital image processing ,indexing system designing, is a Multi-disciplinary crossed research topic and hot spot. Symbol search engine , whose theory is also used in Optical Character Recognition. This technology, has an important academic significance and practical application value.Symbol searching based on the shape , including the search engine dedicated spider, image regional shape feature extraction, index structure, in which the first two technologies are more common point for discussion, but how to design the index system, how to improve the retrieval efficiency is still in the air. The content-based image searching, after retrieved the feature vector, put these features into the database directly. When the number of feature increased, caused the frequently modified in database fields. So a new data indexing system for search engine is needed, and no more relying on ordinary database. This paper introduced the image regional feature extraction algorithm, classified the feature using support vector machine ,based on these theory this paper proposed a indexing system. Firstly, used multi-threads(process) to implement the crawling of symbols on the Internet, used Zernike Moment to extract the shape feature , and stored the local information in feature-inverted indexing system, as retrieving function.This paper mainly introduces the symbol searching system based on shape , and on this basis ,achieved a full set of symbol search engine solutions, including traditional text retrieval and symbol retrieval, also discussed the prospects of Content Based Image Search in the future .