Special Issue Article
Surveillance Mining System for Low Resolution Face Image Recognition Using Kernel Coupling
Video surveillance systems for face recognition are confronted with low-resolution face images. Low resolution face images coming from real time video does not give discriminant information to identify similar images in a dataset. Traditional method solved this problem through employing super- resolution (SR). But these are time-consuming, sophisticated SR algorithms. These algorithm are not suitable for real-time applications. To avoid the limitations, in this work, new feature extraction method for LR faces called coupled kernel distance metric learning (KCDML) is proposed without any SR pre-processing. By using a kernel trick and a specialized locality preserving criterion, we formulated the problem of coupled kernel embedding as an optimization problem whose aims are to search for the pair-wise sample staying as close as possible and to preserve the local structure intrinsic data geometry. Instead of an iterative solution, one single generalized Eigen- decomposition can be leveraged to compute the two transformation matrices for two classifications of data sets.