Face Recognition using Sparse Projection Axes

Recent advances in sparse coding and compressed sensing have paved the way for novel techniques in a variety of fields, including face recognition. Following this trend we present in this paper a feature extraction technique based on projection coefficients computed using a number of sparse projection axes. The feasibility of the technique is demonstrated in a series of face verification experiments performed on the XM2VTS database. The results of our experiments suggest that the proposed technique easily outperforms the popular principal component analysis technique on clean data and exhibit less sensitivity to partial occlusions of the facial images.