Multiple-description coding for robust image watermarking

If we treated image watermarking as a communication problem using the image as the communication channel, applying error-correcting codes (ECC) to the watermarking system should increase the robustness significantly and there has been a lot of works done in this direction. For fading channels, using diversity or signal repetition could be more effective than other ECC methods. However, twice the signal repetition needed twice the bandwidth which can be excessive because the available bandwidth in watermarking was usually small. Multiple-description coding was a mean of trading off between transmission bandwidth and bit error rate. Thereby, we could increase diversity without increasing as much bandwidth as signal repetition. In this paper, we proposed a new image watermarking method using multiple-description coding to increase robustness. Since traditional multiple-description coding was considered in on-off channels where channels were not marred by bit errors but occasional connection outages such as dropped packets, to apply multiple-description coding in watermarking, we needed a form of multiple-description coding for noisy channels instead of on-off channels. Iterative coding of multiple descriptions (ICMD) has been used in noisy channels. Here, we used ICMD together with spread spectrum watermarking to form our watermarking system. We found that bit error did not happen until we compressed the test image in JPEG to 12.7% of the original size (PSNR 37.25 dB). We concluded that ICMD was a very good way to increase robustness for image watermarking.