Human face has a fundamental role in the appearance of the person. Today, the importance of facial surgeries is clear for everyone. Thus, there is a need for the appropriate and accurate facial skin segmentation in order to extract different features. Thus far, many methods have been presented for detecting skin under different illumination conditions. In this paper, an algorithm was proposed which was capable of accurate skin segmentation by modifying varying illumination in color images. The proposed algorithm was robust to determine the accurate border of facial skin in the color images taken from front and side perspectives and its 94.34% accuracy on the tested data indicated its efficiency relative to other algorithms including standard EM algorithm.