彩色图像的快速人脸检测方法研究

Study on Fast Face Detection of Color Image

提出一种快速人脸检测方法。其原理是通过建立肤色模糊集 ,初步确定人脸所在区域 ,然后利用经过充足样本训练后的稀疏神经网络 (SNo W结构 )来验证人脸的存在。为了加快人脸检测速度 ,提出外螺旋搜索策略。实验结果表明 ,该方法在保证较高的检测率的同时 ,大大提高了人脸检测速度 。

A new approach of fast face detection is proposed.The skin detection method to segment the skin-like areas is used.Then a SNoW classifier is used to perform a candidate verification.A multi-scale and helix-like searching strategy is used to accelerate the processing speed.Experimental result shows that this approach is efficient and robust.