Abstract

In this paper, a novel 3D face recognition scheme by Sliding Complex Wavelet Structural SIMilarity (SCW-SSIM) index carried out on detailed geometry images is proposed. First, the preprocessed 3D faces are mapped into geometry images on 2D plane. Then, a multi-scale wavelet transform is designed to decompose geometry image into a set of sub-images and denotes high-pass sub-images of wavelet transform as detailed geometry images, which are utilized as expression-invariant feature for face matching and classifying. At last,similarity metrics obtained from SCW-SSIM work as basis for identification and verification tests. Extensive experiments conducted on FRGC (Face Recognition Grand Challenge) v2.0 show a better performance, which also approve that the proposed approach is promising for dealing with expression distortion in 3D face recognition.

Paper Details