In this paper, a novel method for compressing color images using binary image segmentation and contour polygons is presented. This method encodes the image information by a number of polygons, which represent image regions with identical bit values in the color components. In the color space model RGB for example all neighbored pixels with the same bit value in the color components R, G or B are merged together to one region. This process allows us to represent a color image by number of binary images, which can be described through their contour polygons. The decompression of the image is achieved through reconstruction of the RGB values of every pixel by reversing the compression process.