Quantization of color signals

We develop and evaluate procedures based on subjective criteria for optimizing the design of quantizers for the digital coding of color signals. We start with a series of subjective tests that determine the visibility of quantization noise in several different circumstances. Specifically, two distinct functional relationships for the visibility of quantizing noise in color components are investigated: 1) its dependence on the slope of the luminance signal, 2) its dependence on the slope of the chrominance signal. Using fidelity criteria determined by these visibility functions, quantizers are optimized for the following encoding situations: 1) adaptive PCM coding of the chrominance components based on the luminance slope, 2) element DPCM coding of the individual color components, 3) adaptive DPCM coding of the chrominance components based on the luminance slope, 4) plateau coding of the chrominance components. Pictures are coded using quantizers for encoding situations 2), 3), and 4) Advantages of optimization are brought out by comparing the performance of the optimum quantizers with quantizers optimized on the basis of certain commonly used criteria or with other quantizers. In all cases, we find that for a given bit rate, our optimum quantizers outperform the others in terms of quality, of the coded pictures.