A new median adaptive predictor with applications in subband coding

This paper presents a new adaptive predictor, developed for lossless differential pulse code modulation full-band and subband image coding. The predictor is based on weighted median filter theory, and is compared, using numerical and graphical methods, with a number of alternative predictor designs, both fixed and adaptive. The performance of the predictors is investigated for two different subband coding methods, sub-sampling and pyramid coding, and is found to be higher than of other predictors for a variety of images.