Real Time Face Detection Based on skin tone detector

摘要

Face detection is critical to the application based on human face. Accurate and quick face detection has a determining effect on its following process, like face verification and face recognition etc. Adaboost algorithm [4] has been regarded as a very effective machine learning algorithm that can be applied to the face detection. However, with the increasing of image size determined by the advanced video or image capture device, the search space for finding face is incremented dramatically, to get these images processed in given time is becoming a challenge. In this paper, an algorithm combining skin tone detector and adaboost learning structure is put forward to push the state-of-art of the algorithm, it could reduce the number of window to be scanned to 20% of the conventional approach, thus can satisfy the real-time requirement in our related application where the facial images are captured from video stream of D1 format.