Kernel-Based Pose Invariant Face Recognition
Chao-Kuei Hsieh, Yung‐Chang Chen · 2007
The performance of a face recognition system degrades incredibly due to the variation of facial appearance with different pose, which is well known as one of the bottlenecks in face recognition. One of the possible solutions is generating virtual frontal view from any given non-frontal view to obtain a virtual face. The ideal solution is to reconstruct a 3D model from the input images and synthesize the virtual image with corresponding pose, which might be too complex to be implemented in a real-time application. By formulating this kind of solutions as a nonlinear pose normalization problem, we will propose an algorithm integrating the nonlinearity of kernel function and the efficiency of linear regression method, which modifies the linear assumption in local linear regression (LLR) method and makes the solution more resembling to the ideal one. Some discussions and experiments on CMU PIE database are carried out, and show that our proposed method performs well.