Face Recognition System: Alignment And Illumination Using Sparse Representation

Abstract: face recognition algorithms work on public data sets, but it degrade sharply when they are used in a real recognition system. This is due to the difficulty of illumination, image misalignment in the test image. We consider a fact where the input images are well controlled and test images are only low controlled. We propose a proper simple face recognition system that runs against a high degree of robustness and illumination variation, image misalignment. The system uses tools from sparse representation to align a test face image to a set of database images. We determine how to capture a set of database images with some illumination variation that they span test images taken under illumination. In order to find how our algorithms work under practical testing conditions, we have evaluate a complete face recognition system. Our system can efficiently and effectively recognize faces under different conditions.

Keywords: NN, NS, LDA, SVM