Skin detection is one of the main steps in many image processing systems such as face detection, human identicaton, etc. Since now, many methods are proposed to done it accurately. Most of previous methods have tried to find best match intensity distribution with skin pixels in input image. Experimental Results show that these methods cannot provide accurate results for each kind of human skin colors. In this paper, a two step approach is proposed to solve this problem using color probabilistic distribution estimation technique. The proposed approach consist two steps. In the first step, skin intensity distribution is estimated using some train photos of pure skin. In the second step, the skin areas are detected using Gaussian model and optimal threshold tuning. Single scale retinex technique is used as preprocessing step to increase detection rate. In the result part, the proposed approach is applied on human images and the accuracy rate is computed. The proposed approach can be used for all kinds of skin using train stage which is the main advantages of it. Low sensitivity to impulse noise, low run time complexity, and rotation invariant are another advantages of the proposed approach.