Transform-based watermarking of 3D depth-image-based-rendering images

•A transform-based watermarking algorithm for copyright of 3D images is proposed.•The algorithm uses DWT and SVD transforms to utilize their individual properties.•A watermark is embedded in the center image using a chaos-based embedding scheme.•The watermark is extracted blindly from the center, left-eye, and right-eye views.•High performance results obtained for imperceptibility, robustness, and capacity.

With the continuous advancements in 3D display technology, the need arises for new methods to create and distribute 3D stereoscopic content. Depth-Image-Based Rendering (DIBR) has been recently recognized as an effective method to synthesize 3D content due to its low bandwidth requirement and depth adjustment capability. However, as important it is to generate the DIBR views, it is also important to preserve the copyright of DIBR's views when subjected to illegal manipulation and distribution. In this paper, we propose a blind, imperceptible, and robust 3D DIBR digital image watermarking algorithm based on cascading two mathematical transforms, the Discrete Wavelet Transform (DWT) and the Singular Value Decomposition (SVD). The two transforms provide different, but complementary, levels of robustness against watermarking attacks. Moreover, a watermark is hidden using a chaos-based embedding procedure, and extracted blindly at the receiver side, not only from the center image, but also from the synthesized left-eye and right-eye views. Imperceptibility and robustness of the proposed algorithm are demonstrated using a benchmark set of monoscopic images and their corresponding per-pixel depth map images. Simulation results show that the algorithm is imperceptible, and robust against standard signal processing attacks and DIBR-specific attacks, such as pre-processing of the depth image and adjustment of the baseline distance ratios.