Research On Image Retrieval Technique Based On Content

The technology of content_based image retrieval(CBIR) is a research focus at present,CBIR is on thebasis of image’ content itself,which is extracted by computers,such as color,texture and shape.The systemretrieves the feature vectors consisting of the above imformation from the large image database to satisfythe users’ requirement.CBIR has made some achievements after more than10years, but there are still many theoretical andtechnical problems to be studied and solved. For example, there are some differents between the result ofcolor feature_based image retrieval and the judging of people to vision the color, how to choose ameasurement to closer the human vision is very important; How to solve the problem of extracting theshape feature quickly; How to combine the different image retrieval methods to improve the retrievalefficiency problem. This paper studies the above problems, The color feature vector similarity measureachieved good results in retrieval accuracy by using Hausdorff distance, compared with the traditionalEuclidean distance; In the shape feature extraction,it achieve the purpose of rapid extraction by combiningmoment invariants and Fourier described; In response to a single feature retrieval method, the experimentalresults show that the retrieval results are better than the single feature retrieval by combining the color andshape features.During the study of color space selection and quantification, we choose HSV color space bycomparing of different color space and features, and then we quantify the HSV color model reasonably,extract color histogram as the feature to retrieve.,similarity measure using the Hausdorff distance. Duringthe study of shape feature extraction and matching, through the research and analysis of Moment Invariantsand Fourier Descriptors, a new approach is introduced. First to use the Moment Invariants for the selectionof different images, then determines the similar images after using the Fourier Descriptors, and finally sortsthe graphics sequence, it can greatly improve the efficiency of image retrieval.At last,this paper combineboth color and shape features for image retrieval. We have the color and shape features of a linear weightedsummation, then the similarity of two images is obtained,the weighted value of the ratio between the sizerepresenting the importance of color feature and shape feature. Experimental results show that an integratedsearch feature is better than the individual characteristic.