摘要
从3个方面人手提出了一种抗光照噪声强的人脸肤色检测方法:①从常用的颜色空间中选择出对光照因素稳健的肤色子空间;②提出了基于该子空间的光照自适应的肤色检测方法;③进一步采用小波边缘检测的方法精确定位人脸区域。实验结果表明,该方法对抗灯光、太阳光、荧虹灯光,光的阴影等的影响都有好的效果。
A skin detecting method is presented from three aspects. This method can counteract the illuminating noise very robustly： ①from the common color space, the skin subspace! is selected which doesn＇t influenced by illumination; ②propose a illuminating adaptive detecting skin method based on the suitable subspace; ③accurately locate the face region by using more the wavelet edge detection method. Experimental results show that this method has good effects on opposing the influences of lighting, shine, neon lighting, shadow of light.
出处
《中山大学学报（自然科学版）》
CAS
CSCD
北大核心
2005年第5期1-5,共5页
Acta Scientiarum Naturalium Universitatis Sunyatseni
基金
国家自然科学基金资助项目(60373082)教育部科学技术重点基金资助项目(105134)广东省自然科学基金资助项目(021766)
关键词
人脸检测
肤色模型
光照条件
自适阈值
face detection
skin color model
illumination condition
adaptive threshold
