Skin region detector for real time face detection system

摘要

In this paper we present a real-time skin region detection method for high resolution video on embedded platforms. In the implementation, the region of interest is detected using two algorithms, running in parallel. One algorithm eliminates the background and the other one, detects the skin color patches in the image. To allow images to be processed rapidly on hardware, several techniques were employed to improve the speed and reduce the required memory requirement of the algorithm. Experimental results show that the background subtraction and skin detection are comparable to previously published methods. Moreover, the combination of the two can improve the detection result significantly. The implementation and functional simulation of the modules was based on SystemC and from the VHDL synthesis, the module occupied only 9% of LUTs and 20% BRAM on a Xilinx Spartan-3 XC35000 FPGA.