Approximation for a Color Image

In order to form an approximation to a color image, the image is first transformed to a y-u-v representation. The y component is then segmented using a procedure similar to the monochrome case. The luminance segmentation is also used for the chrominance components (as shown below). This is reasonable, since most of the high frequency information (or region boundary information) is contained in the luminance. Given the segmentation, the luminance and chrominance approximations are generated by filling each region with its corresponding luminance and chrominance mean values.

Color image approximation.