A devised approach to optimize color space transformation for image compression

Abstract

A significant step in color image compression is to apply a color space transformation to the original RGB image to concentrate most of its energy in one plane. Standard JPEG-based compression methods utilize predefined transformations like RGB to YCbCr. In this paper, a devised algorithm is proposed to optimize the transformation for each image. The algorithm iteratively increases the energy concentration by optimizing a novel cost function while limiting the matrix determinant to a bounded region.