Local contrast enhancement for human face recognition in poor lighting conditions

Recognizing human faces in various lighting conditions is quite difficult for a surveillance system. The problem becomes more difficult if face images are taken in extremely high dynamic range scenes. Most of automatic face recognition systems assume the images are taken under well controlled illumination. The face segmentation as well as recognition problem is much simpler under such a constrained condition. However, controlling illumination is not feasible while the surveillance system is installed on locations at will. Without compensating for the effect of uneven illuminants, it is impossible to get a satisfactory recognition result. In this paper, we propose an integrated system that first compensates illuminant effect by local contrast enhancement. Then the enhanced images are fed into a robust face recognition engine which adaptively selects important features and performs classification by support vector machines (SVMs). The experimental result shows that the proposed recognition system is better than recently published literatures with two popular human face image databases.