The Research On The Color-Space Feature-based Of The Image Retrieval Method

The research on technology of content-based image retrieval compromises the research findings of computer vision,Graphics image processing,Image understanding and database.It is a technique for retrieval image on the visual feature (color,texture,shape).Among all the features used in the various indexing algorithms,such as color, texture,and shape,color is the most frequently used signature for indexing,because color is an intuitive cue for image retrieval,and is a very effective low-level feature for indexing.In this paper,we mainly discuss the image color and the space information,discuss the method of extract and express the color-space feature,and design an image retrieval system based on this algorithm.This paper first introduce the method of color-based image retrieval,and expound the choice of color space model,the quantification of color space,the extract and express of the color feature,the similarity measure of the feature and the evaluation of image retrieval.Then,although the color histograms is easy to compute and is insensitive to small changes in viewing positions,However,it does not include any spatial information, and is therefore liable to false positives.In this paper,we suggest to use the multi-CCV mixing location information as the indexing vector.To enhance the efficiency of indexing,we first separate the image into two regions,call core and minor region.After the pretreatment,the multi-CCV mixing location information will be calculated for the each part and store in two indexing vectors,and then the similarity measure of Multi-CCV mixing location information will be carried out.The method of feature extract end express we adopt considers the space information of the color,and also consider the shape of the color.The experimental results show that,this method can achieve good results.