Technology Research, Content-based Image Retrieval

Recently, the developments of the telecommunication and the computer technology and the improvements of the storage technology and the capability of the digital image equipment give us more image resource than ever. And thus the solution of how to locate the proper image quickly and accurately is wanted.The image retrieval based on the content has become the studied hotpot. The simple visualized characterized layers of the image such as colors, textures, shapes and space relations of object are more matured at present and different methods are used to symbolize one of these layers. Image retrieval based on the color feature is a popular method that one falls over oneself for studying, and the method it adopts is the intersecting of color histograms.Based on the understanding of the content image retrieval and related knowledge, this article sets focus on the feature extraction, similarity measure and achievement of the image retrieval technology. The present method of image retrieval based on the color feature is simpler both in feature extraction and in calculation of similarity degree. But it will lose the distribution information of the color in space at the same time. According this deficiency, this article proposes an image retrieval method based on block process, which expresses the image content by combining the color feature and space distribution together, and chooses the HSV color model that can reappearing perfect color feature of human vision and abstracts the histograms of block image as the color feature vector.