Several kernel algorithms have recently been proposed for nonlinear discriminant analysis. However, these methods mainly address the singularity problem in the high dimensional fe...
In this paper a novel non-linear subspace method for face verification is proposed. The problem of face verification is considered as a two-class problem (genuine versus imposto...
We present a comprehensive approach to address three challenging problems in face recognition: modelling faces across multi-views, extracting the non-linear discriminating feature...
Motion blurs confound many computer vision problems. The fluttered shutter (FS) camera [1] tackles the motion deblurring problem by emulating invertible broadband blur kernels. How...
Mika et al. [3] introduce a non-linear formulation of Fisher's linear discriminant, based the now familiar "kernel trick", demonstrating state-of-the-art performanc...