The Research And Implementation Of Image Retrieval Based On Color And Texture

There are more and more image information nowadays with the development of Internet. The content-based image retrieval has become a hot research topic. After reading lots of papers about content-based image retrieval, the author design and develop a retrieval system based on color and texture. The paper discusses the algorithms and process of image retrieval based on color, texture and fusing both feature detailedly, The main work is as followings:As for color-based image retrival, the paper select HSV color space, at the same time the transformation between RGB space and HSV space is presented. In order to reduce the dimension of color feature, the paper use unequal interval quantization to quantize color space. In addition, aimed at the defect that the general color histograms can only get the global information of the image and dose not use the location information, the paper use a layer-cake partition strategy, and combine with cumulative histogram to extract features.As for texture-based image retrival, this paper apply wavelet analysis to texture-based image retrieval. It improves the texture retrieval method a certain extent. Aimed at the defect that the pyramid-structured wavelet transform can't catch medium and high frequency information, which include important texture feature, the paper use the method of extracting texture feature based on wavelet transformation of variable tree structure.No matter color-based image retrieval or texture-based image retrieval, they both have their own defects. If fusing multi image features, it will makes better efficiency. The key of fusinging multi image features is similarity measurement, so the paper present a method of similarity measurement, namely, weighted similarity measurement, and it discusses the algorithm of similarity measurement.Based on the above algorithms, the author design and develop a retrieval system based on color and texture, which can retrieval image by color, texture or fuseing both feature. The experiment result indicate that the method of fusing multi features improve the efficiency of image retrieval.