Local edge patterns for color images: An approach for image indexing and retrieval

Abstract

Local Edge Patterns for Color Images (LEPCI) for image indexing and retrieval are a novel feature extraction technique supplied with this paper. The image converted into RGB and LEPCI encodes the one of a kind OR (XoR) operation between the middle pixel of each coloration plane and its surrounding associates of quantized orientation and gradient values. While neighborhood binary styles (LBP) and local orientation and gradient XoR styles (LOGXoRP) encode the relationship among the gray values of center pixel and its associates, we display that the LEPCI can extract effective texture (facet) features in comparison with LBP and LOGXoRP for color images. The overall performance of the proposed technique is tested by engaging in experiments on Corel-10K databases. The impacts of proposed procedure show advancement as far as their evaluation measures in contrast with LBP, LOGXoRP, and other present systems with particular databases.