Gaussian Mixture Model Based Approach to Color Image Retrieval

Recently, with the explosive growth of the World Wide Web resources of digital color images, much research effort has been devoted to the development of efficient content based image retrieval systems. In this contribution we propose a color indexing method based on the Gaussian Mixture Model of a histogram. The model parameters serve as signatures enabling fast and efficient color image retrieval. We present the results of performed experiments which show that proposed approach is robust to color image distortions introduced by compression artifacts. Therefore, the proposed method is well suited for indexing and retrieval of Internet based collections of color images stored in lossy formats.