Abstract
A systematic study has been made of a high efficiency color image coding system. First, following two (HVC) signal quantization schemes, which were based on the Munsell Renotation Color Space, were estimated : i) perceptual uniform quantization of color space using color differences and ii) coarse quantization of i) using the color reproduction tolerance. Significant features of (HVC) signal are pointed out. Second, based on the quantized (HVC) signal, a color image coding scheme, which regards local features of image, is discussed. Characterized DPCM-VQ coders were adopted to the contour and the flat part of the image separately, where these two parts were separated into each other using the signal features estimated. A satisfactory tharp quality image was obtained by the coding of entropy : 2.4-2.9 [bits/pixel]. The first step of the systematic research of image coding is discussed in this paper. Solving remaining problems and constructing the Picture Quality Scale will allow rapid development of sophisticated color image coding.
