Techniques Of Region-Based Color Image Retrieval

Content-based Image Retrieval is a kind of retrieval that uses content of an image for information search, and it is an important direction in multimedia retrieval field.It's major method is to create feature vector by means of extracting image feature, such as color feature, texture feature, shape feature, space feature and semantic feature, and retrieveing the image by means of measureing the feature vectors. Region-based Image Retrieval is an important direction in Content-based Image Retrieval field. It's major method is to segment the image into several regions using image segmention method, and extract feature for every region, and retrieve the image using single-region matching or multi-region matching method. The technology of Region-Based Image Retrieval is a advanced retrieval method for Content-Based Image Retrieval, because Region-Based Image Retrieval extracts the region feature, then it can describe the object in the iamge, and it can improve the retieval precision if the object can be described effectively.This paper introduces the history and main methods of Content-based Image Retrieval. And we research on color space and the technology of image segmention. We propose a clusting image segnetion method and a region growing image segmention, we proposed a seed automatically selected method for seed selected. using image color feature. We compare the two method by doing experiment. We extract color feature, shape feature and space featrue for every region. In the last, we introduce a multi-region matching method-Integrated Region Matching (IRM), and we improve the method by proposeing our method, and we verify the method's effectiveness by doing expriment.We explore a image segmetion system and a content-based image retrireal system under VC6.0 platform We do a lot of experiment on the system to verify our algorithm proposed int this paper. the results prove our algorithm's efficiency.