基于多视点视频特征的3D-HEVC深度视频信息隐藏算法  

An information hiding algorithm in depth map for 3D-HEVC based on feature of multi-View video

作　　者：左力文 骆挺[1,2] 蒋刚毅[2] 郁梅[2] 高巍[1,2] ZUO Li-wen;LUO Ting;JIANG Gang-yi;YU Mei;GAO Wei(College of Science and Technology, Ningbo University, Ningbo 315211, China;Faculty of Information Science and Engineering, Ningbo University, Ningbo 315211, China)

机构地区：[1]宁波大学科学技术学院,浙江宁波315211 [2]宁波大学信息科学与工程学院,浙江宁波315211

出　　处：《光电子．激光》2018年第5期530-538,共9页Journal of Optoelectronics·Laser

基　　金：国家自然科学基金(61501270;61671258);宁波市自然科学基金(2017A610127);宁波市领军和拔尖人才培养工程择优资助科研项目

摘　　要：为了更安全地传输秘密信息和保护3D视频,提出基于多视点视频特征的3D-HEVC深度视频信息隐藏算法。首先结合彩色视频的纹理特征,考虑深度视频不同区域对绘制视点的质量和编码效率的影响,对深度视频进行区域分割。然后针对不同的区域,采用不同的方式调制最大编码单元(LCU)的QP值嵌入秘密信息。最后,用修改后的QP值进行编码压缩,传输视频信息。实验仿真结果表明,相比于原始HTM13.0,本算法编码重建深度视频绘制视点质量的PSNR平均下降0.0015dB,码率平均增加0.035 15%。本文算法能较好地保证绘制视点的质量,对视频流的码率影响较小,且能实现秘密信息的盲提取。To safely transmit the secret information and protect three-dimensional（3D）video,an information hiding algorithm in depth map for 3D high efficiency video coding（HEVC）based on feature of multi-view video is proposed.Firstly,combining the texture features of the color video and considering the influence of different regions in depth video on the quality and coding efficiency of the virtual viewpoint,the depth video is divided.Then,for different regions,the quantization parameter（QP）value of the maximum encoding unit（LCU）is modulated in the different ways to embed the secret information.Finally,the modified QP value is adopted to encode and compress the 3D video.Experimental results demonstrate that the average decrease of peak signal-to-noise ratio（PSNR）for the rendered virtual view image based reconstructed depth video is 0.0015 dB,and the average bit rate is increased by 0.03515%.The algorithm can guarantee the quality of the rendered virtual viewpoint effectively.It also has little effect on the video stream rate,and the secret information can be extracted blindly.

关 键 词：信息隐藏 深度视频 3D-HEVC 量化参数(QP) 虚拟视点 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]