The Image Search Engine

With the rapid development of Internet,multimedia information on the network is also rapidly increasing,so the needing of download, classification and retrieval has developed.Traditional search engine is not appropriate to do that,so it is necessary for the development of multimedia search engines and databases.In the Internet,more of multimedia information is picture, therefore, the research of picture for download and retrieval has become a hot spot currently.On the basis of traditional general search engine,designing image search engine,adding modules for searching, downloading and retrievalling a new image with improving the structure of crawler. There are some of image attributes, such as the subject of image, or images of gray histogram, or comprehensive value of them which are dealed with in these modules.By calculating the gray histogram of image, then getting distance of the two gray histogram of image histogram. Similarly, access to images and texture color histogram difference histogram margin, and so on, can also obtain images of the other attributes the margin through a Gaussian multiplication method of integration of these attributes the margin and get a different image of the judge Comprehensive value, as image search engine based on the search and retrieval.The introduction of"machine learning"in the semi-supervised learning the concept, to deep the function of image search engine, make it automatically absorb new and similar image through the existing images.Experimental results show that the image search engine can download image which are related with image theme from the Internet, based on the attributes to classify images in the database and the establishment of the relevant index. At the same time, the integrated value which is integrated by attributes of images provides a good reference index for the image retrieval system. In this paper, the image search engines work for further studies to provide a certain value.