A Novel, Efficient, Robust, and Blind Imperceptible 3D Anaglyph Image Watermarking

This paper proposes a novel, robust and efficient blind three-dimensional anaglyph image-watermarking scheme for copyright protection which utilizes the nonsubsampled contourlet transform and principal component analysis for embedding and extracting a binary watermark. The binary watermark is encrypted or scrambled using Arnold transform so as to give security and embedded using principal component analysis into the selected subband of the nonsubsampled contourlet-transformed 3D anaglyph image. Nonsubsampled contourlet transform is used due to its varying properties so as to resist against intentional image processing attacks. Principal component analysis is used in decorrelating the data. The proposed system is tested using Middlebury stereo dataset and experimental results show that the proposed scheme is highly imperceptible and has better performance than the existing anaglyph watermarking schemes in terms of both robustness and imperceptibility.