The Similarity Sort of Single Objects on Monotonous Background

Abstract

In this paper, a simple but efficient similarity sorting technique is proposed by which specifically to classify color images with monotonous background into several similar groups. There is only the color attribute to be considered during similarity retrieval. In order to solved the representing inadequacy of the color attribute to spatial characteristic. The key idea is to partition the images and to find major sub-image from each image by a designated majority function, and finally, using major sub-image as a key with threshold to proceeded image similarity retrieval or classification.