Image Analysis and Retrieval Using Color and Texture Similarity Metrics

Abstract

Image analysis is an extraction of meaningful information from image and retrieval is the process of retrieving the desired image or an identical image from a large collection of images. Structural Similarity Metrics (SSIM) compares local image statistics in corresponding sliding windows in the two images and to pool the result spatially. We are interested in new Structural Texture Similarity Metrics (STSIM) and Color Similarity Metrics (CSIM). Those are centered on an understanding of human visual remark and incorporate a wide range of color and texture section statistics. We have applied separate metrics for the gray scale component of texture that is STSIM and it’s a color composition that is CSIM, which are attributes related to different perceptual dimensions. A major association of this work is to develop a new technique for efficient performance calculation of texture similarity metrics. This should be battered to each specific application. The proposed technique simplifies the testing procedures and increases the chances of gaining subjective results. Here we combine structural texture similarity metrics and color similarity metrics and get appropriate results of texture image retrieval process. Experimental results validate that texture retrieval and compression performance of image, estimation based on the proposed metrics significantly overtakes the performance of existing metrics.