Error resilient coding based on shaping from phase-only transform

In this paper, we propose a novel approach of error correction coding based on the shape preserving characteristic of a phase-only transform. The method creates two kinds of descriptions by phase mapping and pixel interleaving respectively, and synthesizes the original image using the true phases and distorted magnitudes. At the receiving end, when both descriptions are received, a high quality reconstruction can be obtained. On the other hand, if only one description is available, a lower quality version could still be decoded form the received description. Images even though they are badly damaged due to the high error rate during transmission can also be recovered with at least acceptable quality. Simulation results show that the proposed scheme outperforms the other schemes such as pixel interleaving and phase scrambling by about 1.5 - 3 dB in PSNR