Image Retrieval Based On Collaborative Segmentation Technology

With the rapid development of computer technology and multimedia technology, itproduces huge amounts of images in every day. It has become a pressing demand to getImage information from the vast amounts of images. Image retrieval technology arisesat the historic moment. The first image retrieval technology is based on the text retrievaltechnology,this technique can easily lead to inaccurate retrieval results. With theprogress of technology, the content-based image retrieval technology is arised. Itsretrieval results are more accurate. However, if it has the image background interference,the extracted image features can’t accurately describe the image foregroundcharacteristics, this can cause inaccurate image retrieval results.In order to solve this problem, this paper proposes a image retrieval method basedon collaborative segmentation. In this method, we used an improved collaborativesegmentation algorithm based on SIFT Flow and gaussian mixture model(GMM) tosegment image. The segmentation algorithm is mainly using GMM method to model theforeground color distribution of the image to improve the stability and accuracy of thealgorithm. The improved SIFT Flow energy function is used in the algorithm formaking up for the lack of image local characteristics. After getting the foregroundobject of image, and then we use the method based on gray level co-occurrencematrix(GLCM) to extract the texture feature of the foreground object. Finally we dotexture feature matching for calculating the similarity of texture characteristics, thenexport the output according to the similarity from high to low. Because we havesegmented image before extracting texture features, so to eliminate the influence ofbackground image, then the texture features can more accurately reflect thecharacteristics of the foreground object. In the end, we can get more accurate searchresults.