Line-based face recognition under varying pose

Much research in human face recognition involves fronto-parallel face image s, constrained rotations in and out of the plane, and operates under strict imaging conditions such as controlled illumination and limited facial expr essions. Face recognition using multiple views in the viewing sphere is a m ore difficult task since face rotations out of the imaging plane can introd uce occlusion of facial structures. In this paper, we propose a novel image -based face recognition algorithm that uses a set of random rectilinear lin e segments of 2D face image views as the underlying image representation, t ogether with a nearest-neighbor classifier as the line matching scheme. The combination of 1D line segments exploits the inherent coherence in one or more 2D face image views in the viewing sphere. The algorithm achieves high generalization recognition rates for rotations both in and out of the plan e, is robust to scaling, and is compulationally efficient. Results show tha t the classification accuracy of the algorithm is superior compared with be nchmark algorithms and is able to recognize test views in quasi-real-time.