Source-channel coding approach to generate tamper-proof images
Saeed Sarreshtedari, Mohammad Ali Akhaee · 2014
Watemarking the original images to protect them against tampering has recently attracted an overgrowing interest. Recent works in this field offer algorithms that not only localize the tampering, but also recover the original content in the lost area. In this paper, we show that this self-restoration problem can be modeled as a source-channel coding problem. The original image is compressed using an efficient source encoder. The output is then channel coded to be capable of tolerating a certain rate of tampering. At the receiver, decoder reveals the encoder output bit stream if the tampering is below the certain limit. Decoder exploits the location of the erased blocks at the decoding, which are known thanks to the embedded check bits. The output of the source decoder is then used to replace the content of the tampered area. We show that approaching the self-restoration problem from this general viewpoint, the performance is significantly improved comparing to the state of the art schemes, in terms of the quality of watermarked image, quality of the restored content, and tolerable tampering rate.