期刊名称：Journal of Theoretical and Applied Information Technology
印刷版ISSN：1992-8645
电子版ISSN：1817-3195
出版年度：2016
卷号：94
期号：1
出版社：Journal of Theoretical and Applied
摘要：In this paper an integrated method that brings together the image features form color, shape and texture is proposed for content based image retrieval (CBIR). The present paper converts the color image in to HSV color space and derives color histograms. The V color space of the image is divided into non over-lapping region of size 9x9. Each region is sub divided into nine non overlapped sub-regions and a feature vector is derived on each sub region. These feature vectors of each sub region compresses the 9x9 multi-region into 3x3. To derive shape feature, textons are computed on each 2x2 grid and the image is converted into �multi region texton matrix� (MRTM). The gray level co-occurrence matrix (GLCM) features are derived on MRTM and the image retrieval is performed on five categories of Wang database images by combining color histograms and the GLCM features of MRTM. The proposed method of CBIR is compared with GLCM and texton co-occurrence method (TCM) and results indicates the efficacy of the proposed method.
