3D face recognition in the presence of 3D model degradations

Abstract

The problem of 3D face recognition has received a growing interest in the past decades. While proposed approaches have proven their efficiency over renowned databases as FRGC, little work has been conducted on the robustness of such algorithm to the quality of 3D models. In this work, we present a study of the robustness of our 3D face recognition algorithm, namely MS-ELBP+SIFT, to face model degradations. Those degradations include Gaussian noise, decimation, and holes. Degradations are generated on a subset of the FRGC database, hence enabling us to compare the robustness of our approach to them. Results are provided through a comparative study with the baseline ICP method.