The methods of locating the high light area of image, determining the color value on the edge of this high light area and the color rectification are discussed to resolve the problem of color saturation of traffic sign from crossing. ExpectationMaximum of GMM based on symbols is used in MaximumLikelihood Estimation and good classification result is gained. Then a fast algorithm for computing region depth is used to mark the depth of highlight area. A color rectification model is proposed and the experimental results and discussion on the methods are given.