Biometric Recognition of 3D Faces and Expressions

-Biometric recognition is the research field which pursues the association of a person’s identity with the biological characteristics or the behavioral characteristics of that person. Face recognition is preferred by many researchers in biometrics because of its noninvasiveness and its naturalness. In this paper, face recognition using 3D scans is explored. Comparing with 2D face recognition, which uses intensity images to recognize a person, 3D face recognition has the advantage of being independent of environment illumination and subject orientation. However, expression variations of the subjects will change the corresponding 3D scans and thus have a negative impact on 3D face recognition algorithms which assume that the 3D faces are rigid surfaces. In this paper, this issue is also addressed by a proposed framework which incorporates the expressions variation of incoming faces. To test the proposed method, a database containing 30 subjects and 120 3D scans (60 neutral scans and 60 smiling scans) was built. The results proved that incorporating the 3D expression variation into the previous algorithm which treats the face as a rigid surface yields important improvements in the performance of the system. Key-Words: face recognition, biometrics, 2D, 3D, range image, PCA, subspace, SVM, LDA