Sequential scalar quantization of color images

摘要: An efficient algorithm for color image quantization is proposed based on a new vector technique that we call sequential scalar quantization. The components of the 3-D are individually quantized in predetermined sequence. With this technique, palette designed very efficiently, while pixel mapping performed with no computation. To obtain an optimal allocation levels along each coordinate, appeal to asymptotic theory, where number assumed be large. We modify theory suit our application, 1evels typically small. utilize properties human visual system (HVS), luminance-chrominance space. A weighting introduced account greater sensitivity HVS luminance than chrominance errors. spatial activity measure also incorporated reflect increased errors smooth regions. yields high-quality images and significantly faster existing algorithms.