一种可变光照条件下的肤色检测算法

A Skin Color Detecting Algorithm Under Variant Light Condition

肤色是人体表面最为显著的特征之一,可应用于人脸检测与跟踪、手势识别等,肤色检测算法在其中居于关键地位。该文提出了一种可变光照条件下的肤色检测算法,首先在YCbCr颜色空间从大量肤色样本中统计出肤色模型,考虑了在高亮度区和低亮度区肤色色度(Cb,Cr分量)与亮度(Y分量)非线性相关;其次,使用一种彩色均衡技术消除由光照引起的色彩偏移。该算法可适用于较广范围的光照条件,可应用于复杂环境下的人脸检测、手势识别等。

As one of the most notable features in the surface of human body,skin color can be applied in the computer vision field,such as face detection and tracking,hand gesture recognition and so on.The skin color detecting algorithm plays an important role in its application.A skin color detecting algorithm under variant light condition is presented.Firstly,taking into account the nonlinerity between chroma (Cb,Cr component )and luminance(Y component )of skin color in the high and low luminance region,skin color model is constructed from a great amount of skin color samples in the color space of YCbCr .Secondly,a color balance technique is used to remove the color bias produced by light condition.This algorithm is fit under wider light condition and can be applied in face detection,hand gesture recognition under complex environments.