A typical method of improving the contrast of images is HE(Histogram Equalization). This method makes uniform distribution by adjusting the histogram[1]. However, this method is not concerned about the human visual system so as to cause visual fatigue in stereo image. In this paper, we propose a method for distinguishing object from background in image based on Roberts edge detection. The experiment result validates that our proposed algorithm is reliable to achieve brightness contrast ratio of 30% between main object and background image. Based on this adjustment on brightness contrast ratio, our stereo image can obtain a more comfortable stereo image with a less visual fatigue.