A Method of Face Detection Based on Skin Color Model in Fixed Scene

This paper presents a rapid detection method for face detection in stationary scenes. First of all, we obtain the binary image based on background picture subtraction and extract the moving target area with SHEN filter. Then, we select the YCbCr color model, based on skin-color clustering model, we can do color segmentation on the sub-image of moving target. Following that, we use elliptical template to detecting the face region and marking the face positioning. Lastly, we implement the algorithm with Matlab7.0, and do experiments in our face testing set. The experimental data manifest the good robustness of the suggested method for face detection in the stationary scenes, and its strong adaptability to attitude, expression and age. Also, the results show that this method is definitely practical in real time. In a word, the accuracy is above 84 percent.