Face recognition using weber local circle gradient pattern method

Abstract

The Weber Local Descriptor (WLD) is a classical and efficient face representation method, but it has the shortage that it employs the contrast information between the center pixel and its eight nearest pixels, which can also be sensitive to illumination variation. In order to overcome the shortcomings mentioned above and solve the problem of sensitivity to the illumination, we propose a novel face recognition algorithm, Weber Local Circle Gradient Pattern (WLCGP), which not only takes the relationship between the target pixel and the surrounding pixels into account, but also considers the relationship among the surrounding pixels. Through calculating the overall gradient information and the cycle gradient information of an image, the WLCGP method can produce the fusion characteristic and extract more effective and discriminative feature information. Finally, we demonstrate the superiority of the proposed WLCGP method over the traditional methods on the ORL, AR face database and the Singapore infrared face database.