Image Error Concealment Based on Watermarking

Imperfect transmission of block-coded images often causes lost blocks. These blocks may contain very important information of image. In this paper a new error concealment scheme is proposed for restoring lost blocks of images. It is based on the robust watermark technique to restore lost blocks in the region-of-interest (ROI) and the best-neighborhood-matching (BNM) technique to restore lost blocks in the region-of-background (ROB) of images. The ROI of image contains the human-interested information and is the most important part of image. It can be defined by user or decided by some algorithms. In addition, a novel fragile watermark technique is designed and applied to detect the error blocks of images. With this technique, the proposed scheme can find out all the errors of damaged images. Experiments show that very good restoration results are obtained by the proposed scheme, especially for the human-interested regions of images.