A PERCEPTUALLY OPTIMIZED JPEG-LS CODER FOR COLOR IMAGES

摘要

Perceptual coding has long been recognized as a promising approach to achieve high efficiency in compressing digital images, since coding bits can be saved from representing the part of the image that is not perceivable by the human visual system (HVS). The perceptual redundancy inherent in color images can be found on the fact that the HVS cannot discriminate color signals of small differences. Color images are therefore expected to be encoded more efficiently if coding distortion can be shaped as the major part of perceptual redundancy. In this paper, the JPEG-LS coder in the near-lossless compression mode is perceptually optimized by making coding errors imperceptible or minimally noticeable. A visual model is introduced to estimate the thresholds of just noticeable distortion (JND) for three color channels of each pixel. With the estimated JNDs, the coding mode of each pixel in each color channel can be appropriately determined, and the quantization stepsize in predictive coding mode can be perceptually tuned for visually lossless encoding. Simulation results show that the performance of the perceptually optimized coder is superior to that of the un-optimized coder in terms of the bit rate required for achieving the same visual quality.