Face detection and tracking algorithm in video images with complex background

Abstract

The problem in accurate and fast face detection and tracking in video images with complex background is studied in this paper. In order to meet both of speed and accuracy, a face detection method integrating the skin color segmentation and Split up Sparse Network of Winnows (SNoW) is presented. This kind of method can not only utilize the advantage of skin color segmentation to remove the interference of complex background, but also has the advantage of Split up SNoW in accurately classifying feathers. More importantly, this method can inherit their common advantage of fast speed. Based on these advantages, the method is used to detect human face areas and then use the way of Continuously Adaptive Mean Shift (CAMShift) to track human face areas. The experimental results show that the proposed method integrating the skin color segmentation and Split up SNoW is superior to each single method. Through this method we can achieve higher recognition accuracy and faster speed in face detection and tracking.