A new fuzzy adaptive perceptual image coding system

A new adaptive, perceptual image coding system in the framework of fuzzy logic is proposed. The image is decomposed into 8/spl times/8 blocks that are then subjected to the discrete cosine transform (DCT). The image DCT blocks are then classified using the fuzzy c-medians (FCMED) clustering algorithm into four fuzzy classes. Coding bits are then adaptively and perceptually allocated to the DCT coefficients in each class. DCT coefficients are then normalized, optimally quantized, coded, and transmitted, accompanied by their overhead information over a channel. The performance of the proposed system is evaluated using the peak signal-to-noise ratio (PSNR) measure and a just noticeable distortion (JND) based perceptual measure.