High-compression of chrominance data by use of segmentation of luminance

The paper describes a very simple technique for compression of chrominance data in colour images and video sequences coded at low bit rates. The technique is based on coding of arbitrarily shaped regions, whereby each region is efficiently represented by one value of chrominance which is averaged over the whole region. As the regions are determined on the basis of reconstructed luminance component, no information upon the segment shapes has to be transmitted. In order to obtain high compression, the set of assigned pairs of chrominance is processed by vector quantization. Experimental results prove high efficiency of the proposed technique.