Abstract: In the analysis of textured images using fractal dimension to evaluate the state of human skin, we applied the power spectrum method to estimate the fractal dimension of images of human skin surfaces. We extracted features from these skin images to classify the age and gender. For a good estimation, achieving pictures that include accurate information among taken pictures is necessary. But because of artifact exist in skin image; sometime calculate an accurate fractal dimension is impossible. In addition fractal dimensions calculated from different pictures of the same skin are different together. So in this paper we propose an effective algorithm to skip over images include artifact, and calculate fractal dimension from images without any artifact.