云计算环境下基于立体视觉的3D高清视频信息隐藏研究

三维(3D)高清视频由于能够提供立体感和交互性的全新视觉体验而受到人们的欢迎，然而，在云环境下，3D高清视频的私密性和安全性受到较大的挑战。加密彩色视频，同时将秘密信息嵌入深度图序列，使云端管理者在不知视频内容的情况下，保护3D高清视频的版权和内容完整性。基于3D-HEVC的彩色视频加密已有较成熟应用，但是基于立体视觉感知的3D高清视频信息隐藏有待深入研究：(1)根据人类视觉感知特征，通过主观实验分析影响深度失真掩蔽特性的因素，建立指导深度图序列信息隐藏的深度失真掩蔽模型；(2)结合3D-HEVC的深度图序列编码过程的信息嵌入算法；(3)基于深度失真掩蔽模型，根据多视点深度图协调机制以及感兴趣区域计算模型，合理分配信息隐藏容量，并建立深度失真与信息隐藏容量的关系模型，从而取得最优化的信息隐藏性能。本项目将为3D高清视频信息隐藏提供方法和理论依据，在多媒体技术与网络通讯领域具有重要应用价值。

Since three dimensional (3D) high definition video provides binocular vision, and novel interactive visual experience, it is welcomed by people. In the cloud environment, privacy and security of 3D high definition video are challenged. Color video is encrypted, and meanwhile, the secret information is embedded into the depth map sequence for allowing cloud administrators to protect copyright and ensure content integrity without knowing the video content. Based on 3D-HEVC, video encryption application has been mature, but binocular vision based depth map sequence data hiding needs to be further studied: (1) according to characteristics of the human visual perception, depth distortion masking characteristics are analyzed by subjective experiments, so that its masking model is built for depth map sequence data hiding; (2) Basic theory of data embedding combining encoding processes of 3D-HEVC based the information depth map sequence; (3) Based on depth distortion masking model, using coordination mechanism of multiple depth maps and computation of region of interest, so that data hiding capacity for different areas is allocated properly. Moreover, relationship between depth map distortion and hiding capacity is built for achieving optimized performance of data hiding. This project provides methods and theories of 3D high definition video data hiding, and has important applications in the multimedia technology, network communications.