An Integrated CBIR Scheme Based on Distance Transform

Abstract

In this paper, we developed a new set of content-based descriptors that incorporated three primitive features of color images: chromatic, spatial, and spectral. Based on the distance transform, the proposed local relative distances pattern (LRDP) records the distance pattern of a color plane. Thus, LRDP can be regarded as an integration of chromatic and spatial information. The spectral information for the boundary and texture portion of an image is extracted by the devised frequency histogram. These two descriptors are integrated to form a robust CBIR system. Experimental results show that the proposed scheme has better retrieval accuracy than conventional CBIR methods