Skin detection is an important work before face detection and tracking, hand detecting, adult image filtering, etc. But it is very difficult in images with complex background. Some non-skin pixels are usually mistakenly detected as skin pixels. This paper proposes an improved skin detection method that combines RGB color space and YCgCb color space, then uses the texture information detect skin areas. The experiment results show that this method can separate skin and non-skin color under complex background, it has higher detection rate.