Multi-View Face Detection Based on AdaBoost and Skin Color

This paper describes a multi-view face detection method which combined skin color information and AdaBoost-based face detection technique together to improve the detection accuracy and detection speed. First the input image is converted into YCbCr color space, then the image is binarized according to skin threshold and a binary image was obtained in which pixels with value 1 are skin-pixels and pixels with value 0 are non-skin pixels. After that only the skin regions are scanned by multi-AdaBoost detectors which were also trained on skin binarized face images. Before scanned by the multi-AdaBoost detectors, the sub windows are filtered by the ratio of the skin- pixels in the sub windows which can eliminate most of the false face regions .The experimental results show that the proposed method can achieve high detection accuracy with fast detection speed.