Efficient scarf detection prior to face recognition

Abstract

Face occlusion is a very challenging problem in face recognition. The performance of face recognition system can decrease drastically due to the presence of partial occlusion on the face. One approach to overcome this problem is to first pre-classify faces into two classes: the clean face and the occluded face; then faces in different classes are treated by different recognition systems. In this case an algorithm which is able to automatically detect the presence of occlusions on the face will be a useful tool to increase the performances of the system. In this paper we present a scarf detection algorithm. In the experimental results the performances of the algorithm are reported and compared with state of the art systems.