Implementation of face detection algorithm based on KL-Gaussian Model on DSP

Abstract

Human face detection plays an important role in applications such as secure access, video surveillance. In view of the poor detection rate and low speed of the detection for faces in profile view, pose varied and complex background, this paper proposes the KL-Gaussian Model algorithm in YCbCr color space for face detection, which is designed and implemented on the TMS320DM6437 platform. Furthermore, to develop the DSP platform advantage, the optimization scheme for our designed face detection system is proposed. The result shows that the performance of the system is good with high detection rate and good robustness in the condition that faces with different poses and angles, which is significant to the intelligent development of video surveillance system.