The face images fusion based on Laplacian pyramid and LBP operator

Abstract

Contraposing specialty of face texture feature, we present a novel multi-resolution fusion method for visible and near infrared (NIR) images based on Laplacian pyramid and Local Binary Pattern (LBP). As we know, the visible images have sufficient texture and specific information and NIR images are illumination invariant. Therefore, we fuse a couple images which include one visible image and one NIR image to obtain a composite image with more texture information and less influence by variation of illumination. The Local Binary Pattern (LBP) operator, as a simple and efficient face recognition method, is a powerful and attractive texture descriptor. We utilize it to describe the different regions from visible and NIR images and decide the method of fusion to be used. The comparison with the other fusion rules proves the effectiveness of the proposed method on visible and NIR face images.