Integration of Color and Local Derivative Pattern Features for Content-Based Image Indexing and Retrieval

Abstract

This paper presents two new feature descriptors for content based image retrieval (CBIR) application. The proposed two descriptors are named as color local derivative patterns (CLDP) and inter color local derivative pattern (ICLDP). In order to reduce the computational complexity the uniform patterns are applied to both CLDP and ICLDP. Further, uniform CLDP (CLDPu2) and uniform ICLDP (ICLDPu2) are generated respectively. The proposed descriptors are able to exploit individual (R, G and B) spectral channel information and co-relating pair (RG, GB, BR, etc.) of spectral channel information. The retrieval performances of the proposed descriptors (CLDP and ICLDP) are tested by conducting two experiments on Corel-5000 and Corel-10000 benchmark databases. The results after investigation show a significant improvement in terms of precision, average retrieval precision (ARP), recall and average retrieval rate (ARR) as compared to local binary patterns (LBP), local derivative patterns (LDP) and other state-of-the-art techniques for image retrieval.