Disparity map based watermarking for 3D-images

This paper proposes a watermarking system that is capable of surviving the reconstruction phase of disparity-based image rendering. When the hole (missed pixel) detection and disparity maps are combined, an intermediary matrix is created. This matrix in turn is used to generate a decision matrix. This information helps limit the watermark embedding process to regions that survive the rendering process, while minimizing detection of false positives. Through the tests that were run no Bit Error Rate (BER) change was observed between the original and rendered images. The average Peak Signal to Noise Ratio (PSNR) and the average storage capacity were 36.2 dB and 62.5%, respectively.