Date of Award
Master of Science
Mongi A. Abidi
Andreas Koschan, Seong G. Kong
This thesis presents a novel illuminant invariant approach to recognize the identity of an individual from his 3D facial scan in any pose, by matching it with a set of frontal models stored in the gallery. In view of today’s security concerns, 3D face reconstruction and recognition has gained a significant position in computer vision research. The non intrusive nature of facial data acquisition makes face recognition one of the most popular approaches for biometrics-based identity recognition. Depth information of a 3D face can be used to solve the problems of illumination and pose variation associated with face recognition.
The proposed method makes use of 3D geometric (point sets) face representations for recognizing faces. The use of 3D point sets to represent human faces in lieu of 2D texture makes this method robust to changes in illumination and pose. The method first automatically registers facial point-sets of the probe with the gallery models through a criterion based on Gaussian force fields. The registration method defines a simple energy function, which is always differentiable and convex in a large neighborhood of the alignment parameters; allowing for the use of powerful standard optimization techniques. The new method overcomes the necessity of close initialization and converges in much less iterations as compared to the Iterative Closest Point algorithm. The use of an optimization method, the Fast Gauss Transform, allows a considerable reduction in the computational complexity of the registration algorithm. Recognition is then performed by using the robust similarity score generated by registering 3D point sets of faces. Our approach has been tested on a large database of 85 individuals with 521 scans at different poses, where the gallery and the probe images have been acquired at significantly different times. The results show the potential of our approach toward a fully pose and illumination invariant system. Our method can be successfully used as a potential biometric system in various applications such as mug shot matching, user verification and access control, and enhanced human computer interaction.
Ayyagari, Venkat Rao, "Pose Invariant 3D Face Authentication based on Gaussian Fields Approach. " Master's Thesis, University of Tennessee, 2005.