Adaptive predictor based on maximally flat halfband filter in lifting scheme

For complex short time-varying signals, a high-order predictor does not always yield good performance. For this, we investigate the use of a short-order adaptive predictor. Since the maximally flat filters are the optimal predictors for polynomial signal prediction, the adaptation is based on the combination of a set of maximally flat filters. For compression efficiency, the dynamic ranges of the weighting variables are specially considered. For this, based on the Bernstein filters, another form to represent the weighting variables is used. These two sets of weighting coefficients can be transformed into each other with a simple linear transform. Thus, the adaptation can be made in both the time domain and the frequency domain. For block-based image coding, the least square criterion is used to derive the weighting coefficients. Experimental results show that the adaptive predictor performs better than the S+P transform, the median edge detector (MED), and the gradient adjusted predictor (GAP)