Local Smoothness Pattern for content based image retrieval

Local binary pattern (LBP), which efficiently summarizes the local structure of images has been very successful in image retrieval applications. In this paper, we propose a Local Smoothness Pattern (LSP) for image retrieval applications which captures the smoothness over the curve directions in a local region. To compare its performance, this technique is applied to a database of 1180 sub-images from the twenty different classes of Brodatz database and is compared with the LBP. Result shows that the proposed technique achieves good retrieval rate.