Differential encoding of high-resolution data

Predictive models for generating side information in differential encoding of data (e.g., images) sampled at high densities are considered. A model based on incremental difference operators is used in place of the conventional linear predictive coding (LPC) paradigm. It is seen that, unlike standard LPC, this formulation is stable for high sampling densities. Moreover, this formulation leads to more efficiently encoded side information in this regime.