Different environment illumination has a great impact on face detection and recognition. The automatic detection and radiation correction of a highlighted face area is helpful to analyze and identify human faces correctly in a color image. In this paper, a novel approach is presented based on a dichromatic reflection model to detect and correct highlighted skin pixels in the TSL(tint-saturation-luminance) color space. After inspecting the distribution configuration of skin pixels in various color spaces and the spectrum reflection features of human skin, the authors carry out the highlighted are a analysis on a critical two-dimensional plane instead of in a three-dimensional color space, which brings forth certain advantages: computation complexity is deduced, rates between eigenvalues are produced in stepwise PCA to detect automatically the existence of highlighted skin and estimate robustly the skin dichromatic reflection vectors. The highlighted skin regions are compensated based on the dichromatic reflection model.