Two-dimensional hybrid image coding and transmission over noisy channels

The image coding performance of a two-dimensional (2D) hybrid system when transmission occurs over a noisy channel is described. Image data compression is achieved through the use of 2D transform operations on the gray-level pixels within an image subsection, followed by 2D DPCM (differential pulse code modulation) encoding of common transform coefficients from disjoint image subsections. Results from this 2D hybrid scheme indicate an overall improvement in image source coding gain vis-a-vis conventional 2D transform coding. An adaptive hybrid scheme is investigated in which both the quantizer characteristics and the predictor structure are tailored to fit the statistics of the transform samples in each disjoint block. Hybrid system performance dependence on block size and choice of transform is demonstrated. Two-dimensional hybrid image coding and transmission over noisy channels is investigated analytically for Gaussian autoregressive image models and experimentally through computer simulation on real-world images. This hybrid system provides a means of improving existing transform coding systems with a minimal increase in hardware complexity