CBIR based on singular value decomposition for non-overlapping blocks

Abstract

Image retrieval techniques are useful in many image processing applications. Conventionally, the content based image retrieval systems work with whole images and searching is based on comparison of the query and database images. The retrieval of the images is commonly based on the color, shape, texture and others. However, the color only or texture only or shape only cannot enough to discriminate the image. Therefore, in this paper, we present a proposed scheme based on the color and texture information using the Singular Value Decomposition feature to achieve better results. The proposed scheme consists the process of converting an image to HSV color space and each color plane is partitioned into 4, 16, 64 non-overlapping blocks. Later, a one to one matching procedure is implemented to compare the query and target images by using the Euclidean distance for retrieving the similar images. Experimental results demonstrate that the proposed method has higher retrieval precision than other conventional methods color moments, Dominant color, Dominant color and GLCM texture methods. Also shows that SVD features represent the color and texture features of the image