Skip to Main Content
In this paper, a novel method based on Gabor Wavelet Transform plus Local Binary Pattern is proposed for feature representation in 3D face recognition. Firstly, to enable the application of the new feature representation of the framework, Iterative Closet Point (ICP) is utilized as preprocessing to detect the nose tip by aligning a pair of symmetrical 3D face models of the same person, crop the interest of region, and then generate the corresponding depth images. Secondly, multi-scales and multi-orientations Gabor Wavelet Transform is proposed for feature representation. Then, LBP operator is used to describe the local features future. For each face region, the statistical histogram is utilized to represent the 3D face accordingly. Finally, LSDA based strategy is presented to address the recognition task. The new algorithm benefits mostly from two aspects: One aspect is that Gabor wavelets are promoted for their useful properties, such as invariance to illumination, rotation, scale and translations, in feature extraction. The other is that the LBP operator not only describe the local features which have been demonstrated to be quite considerable in recognition task, but also reduces more coefficients for image representation. Experiments based on the FRGC 3D face database demonstrate the effectiveness and efficiency of the new method. Results show that our new algorithm outperforms the other popular approaches reported in the literature and achieves a much higher verification rate at 0.1 FAR.
Date of Conference: 3-5 Aug. 2011