基於圖片描述子之三維影像數位浮水印

Image Descriptors Based Digital Blind Watermarking for DIBR 3D Images

繆昕(Hsin Miao)

指導教授 ： 吳家麟

國立臺灣大學/電機資訊學院/資訊工程學研究所/碩士(2013年)

https://doi.org/10.6342/NTU.2013.01487

摘要

隨著三維數位多媒體內容（如三維立體電影）的普及，如何保護其智慧財產權為一重要之議題。現今三維影像普遍的資料表示形式為彩色圖像搭配其對應之深度圖，並利用深度圖生成技術（DIBR）產生立體之影像；我們提出的數位浮水印架構便是適用於保護此類型之三維影像。此系統運用圖片描述子（image descriptors）輔助，以重新校準（resynchronization）的方法還原DIBR產生立體之影像，至原先放置數位浮水印之影像，接著即可在還原後的影像偵測數位浮水印。與前人之系統相比，我們提出的系統可以保護由MPEG組織提出的DIBR生成之三維影像。此外，我們亦針對不同的圖片描述子（SIFT以及CHoG）比較；並對針對常見的壓縮攻擊（JPEG以及HEVC）探討此數位浮水印技術在此攻擊之下的穩定性。

關鍵字

三維數位多媒體 ； 深度圖生成技術 ； 數位浮水印 ； 圖片描述子 ； 重新校準

並列摘要

Content protection for 3D multimedia data is essential to assure property rights. The depth-image-based rendering (DIBR) operation is one of the ways to synthesize arbitrary virtual views from color-plus-depth 3D data. In this work, a novel blind watermarking scheme is proposed to protect DIBR 3D images. The watermarking system utilizes image descriptors as side information to compensate the distortion produced by DIBR operations. The compensation process, which is named as resynchronization, estimates the disparity map between the views, and recovers the synthesized virtual view back to the watermark embedded view. Compared to the previous work, the proposed method is able to detect embedded watermark on arbitrary DIBR synthesized virtual views, defined in MPEG standard. Also, we find that the side information reduction process based on SIFT descriptor performs better than that of another compressed descriptor, CHoG. Furthermore, simulation results show that the proposed scheme is robust against JPEG compression. Finally, the robustness of our work against HEVC (H.265) 3D compression is also investigated.

並列關鍵字

Blind Digital Watermarking ； Image Descriptor ； 3D ； Depth-Image-Based Rendering ； Resynchronization

參考文獻

[2] Call for Proposals on 3D Video Coding Technology. ISO/IEC JTC1/SC29/WG11 MPEG2011/N12036, Mar. 2011.

[3] Test Model under Consideration for HEVC based 3D video coding. ISO/IEC JTC1/SC29/WG11 MPEG2011/M12350, Nov. 2011.

[4] P. Bas, J.-M. Chassery, and B. Macq. Geometrically invariant watermarking using feature points. Image Processing, IEEE Transactions on, 11(9):1014–1028, 2002.

[5] V. Chandrasekhar, G. Takacs, D. Chen, S. Tsai, Y. Reznik, R. Grzeszczuk, and B. Girod. Compressed histogram of gradients: A low-bitrate descriptor. International Journal of Computer Vision, 96(3):384–399, 2012.

[6] C. Deng, X. Gao, X. Li, and D. Tao. Local histogram based geometric invariant image watermarking. Signal Processing, 90(12):3256 – 3264, 2010.