PROJECT TITLE :

A Robust Group-Sparse Representation Variational Method With Applications to Face Recognition

ABSTRACT:

For face recognition applications, we offer a Group-Sparse Representation-based technique (GSR-FR). A non-convex penalty for sparsity and a resilient non-convex loss function are included in the new sparse representation variational model. By employing an approximation of the l 0 quadsinorm, the penalty pushes groups to be sparse, and the loss function is chosen to be resilient to noise, occlusion, and disguise. Nonconvex optimization problems can be easily solved using a majorization-minimization technique paired with forward-backward splitting, which in particular reduces the solution to a sequence of easier convex optimization sub-problems that are more manageable. To verify the GSR-FR algorithm's viability, we conducted extensive experiments on a wide range of face databases and found it to be competitive with current methods that use sparse representation.