随着多媒体通信及网络技术的飞速发展，人们对视频应用的需求日益广泛。由于Internet固有的网络异构性、用户终端设备显示能力、处理能力和显示分辨率也存在明显差异，这对视频编码技术提出了更高的要求，视频编码应具有一定的灵活性和可伸缩性，即能够动态地根据网络状态、设备性能及用户需求提供不同帧率、分辨率和码率的视频。可伸缩视频编码(Scalable Video Coding，SVC)因其能够较好地解决上述问题而得到广泛关注。近年来，基于三维小波的可伸缩视频编码成为视频编码领域的研究热点。本文在对已有的三维小波可伸缩视频编码框架进行分析的基础上，根据三维离散小波变换(Discrete Wavelet Transfotin，DWT)的特点，针对目前三维DWT的可伸缩视频编码的若干关键问题展开研究。具体地，本论文的主要研究工作如下： (1)基于率失真优化准则的自适应Lagrange乘子选择技术 本文分析了在三维DWT可伸缩视频编解码系统中，Lagrange乘子作为一个关键的编码参数，Lagrange乘子的选择问题会接影响到视频编解码系统码率失真性能和编码效率的好坏。本文针对现有Lagrange选择技术存在的不足，通过对小波域运动补偿时域滤波(Motion Compensated Temporla Filtering，MCTF)技术的研究，在充分考虑小波滤波器类型、子带耦合现象和时域子带内容特性这些因素的影响下，提出了基于率失真优化准则的自适应Lagrange乘子选择技术。实验结果表明，该技术不仅可以精确地获得MCTF的误差传递模型，提升MCTF的效率，还可以更好地编码时.空域子带内容信息，获得平稳、最优的视频重建质量； (2)基于率失真优化准则的λ域码率控制技术 本文分析了在三维DWT可伸缩视频编解码系统中，如果采用传统的基于R-Q (Rate-Quantization)模型的码率控制技术，则存在量化参数是以内嵌方式隐式存在于位平面截取阶段而不是显式表达值，量化步长的取值不能获得的问题。本文针对传统小波域码率控制效果不佳的问题，根据相邻子带之间的时空相关性、子带活动性，基于率失真编码理论分析，提出了可以精确刻画率失真关系的R-λ模型，并提出了基于率失真优化准则的λ域码率控制技术。实验结果表明，与传统的码率控制技术相比，本文提出的λ域码率控制技术可以在精确的码率控制下显著提升率失真性能； (3)基于率失真优化准则的最优编码模式选择技术 本文分析了在三维DWT可伸缩视频编解码系统中，编码模式的选择对视频编码性能和编码效率的影响。在进行最优编码模式选择时，本文针对原有编码模式选择算法会导致编码复杂度过高的缺点，根据自适应小波分解级选择、小波子带像素的连接特性以及人眼感知特性，提出了基于率失真优化准则的最优编码模式选择技术，通过最小化编码代价函数值来选择最优的编码模式。实验结果表明，本文提出的最优编码模式选择技术不仅能够有效提升三维小波可伸缩视频编码的率失真性能、实现码率约束下的最优编码效率，同时还能够使得重建视频质量更加符合人的主观感受；且计算复杂度小，可用于实际编码环境； (4)自适应重叠块运动补偿权重矩阵设计技术 本文分析了在三维DWT可伸缩视频编解码系统中，二维DWT是针对MCTF后的整个时域子带帧系数进行变换，而不像DCT是针对宏块内残差进行变换，因而不同宏块的运动估计的差异性会使得图像宏块边界之间产生比DCT更为严重的方块效应。在进行去块效应时，本文针对传统重叠块运动补偿权重矩阵存在的问题，为克服固定不变的权重矩阵会导致块边缘信息被模糊或被平滑过渡的缺点，提出了自适应重叠块运动补偿权重矩阵技术。所提方法通过确定权重系数矩阵中当前块重要位置像素点的权重系数后，当前块其它位置像素点的权重系数由对角线内插法得到，镜像对称得到相邻块的权重系数。大量实验表明，经过对角线内插法得到的权重系数是平滑且连续的。在进行可变块运动补偿时，自适应权重矩阵是适合于可变块大小的权重矩阵，每一个块类型都有与其对应的权重矩阵。实验结果表明，该技术不仅能够使得图像主观质量显著提高，还能提高编码效率，编码时间平均节省14.81％。 关键词：视频编码；三维小波变换；率失真优化；Lagrange乘子；重叠块运动补偿 论文类型：应用基础

 With the rapid development of multimedia technology, communication technology, and network technology, multimedia applications have been pervasive in people's daily lives. Due to heterogeneous networks and end-userterminal capabilities, the coded bitstream should provide the temporal, spatial and SNR scalabilities to meet the requirements of the clients with diverse display resolutions, bandwidths, computational capability and memory capabilities. Scalable video coding (SVC) has drawn widely concern, because it can solve the problems described above much better. Recently, three-dimensional (3-D) DWT-based SVC has been a hot topic in the field of video coding. Based on the anlaysis the framework of 3-D wavelet-based SVC, with the property of characteristics of discrete wavelet transform, the research of this thesis focuses on the key techniques of 3-D wavelet-based SVC. Specifically, the content of this thesis can be summarized as follows: (1)Rate distortion optimization-based adaptive Lagrange multiplier selection Rate distortion optimization (RDO) plays an essential role in substantially enhancing the coding efficiency. Among all the possible coding modes, it aims to select the one which has the best trade-off between bitrate and compression distortion. Specifically, this tradeoff is tuned through the choice of the Lagrange multiplier. In this paper, an efficient content adaptive Lagrange multiplier selection algorithm is proposed in the context of RDO for 3-D wavelet-based SVC targeting quality scalability. First, we introduce a novel weighting method, which takes account of the mutual information, gradient per pixel, and texture homogeneity to measure the temporal subband characteristics after applying the motion-compensated temporal filtering (MCTF) technique. Second, based on the proposed subband weighting factor model, we derive the optimal Lagrange multiplier. Experimental results demonstrate that the proposed algorithm enables more satisfactory video quality with negligible additional computational complexity; (2)Rate distortion optimization-based λ -domain rate control This thesis proposes a λ-domain rate control algorithm based on RDO technique. After analyzing the difficulties of applying the traditional R-Q (Rate-Quantization) model based rate control algorithm to 3-D wavelet-based scalable video coding, this dissertation builds an R-λ model, which better reflects the rate-distortion relationship. Furthermore, the dissertation proposes an R-λ model based rate control algorithm, and applies it to the latest 3-D wavelet-based video coding. Compared with R-Q model based rate control algorithm, R-λ model based rate control algorithm not only achieves both better rate control accuracy and better rate-distortion performance, but also achieves more accurate instant bitrate; (3)Rate distortion optimization-based best mode decision Currently, rate-distortion optimized mode decision is widely used in scalable video coding. In this thesis, we have fully considered adaptive wavlet decomposition level selection, subband pixel connectivity and human perception, we propose RDO-based mode decision. We choose the mode with the minimum of Lagrange cost as the best mode.Experimental results show that SSIM optimal RDO mode decision encodes image structural information better and gets higher subjective quliaty compared with objective quality based rate control, and has low complexity and thus can be used in actual video coding applications; (4)Overlapped block motion compensation using adaptive weighting matrices design Overlapped block motion compensation (OBMC) is considered to be a key component of most efficient block-based motion estimation and compensation technologies for 3D wavelet-based SVC. However, the conventional OBMC approach globally smoothes out the prediction intensity field with the uniform window function for all the same-sized blocks, with the result that sharp features such as blocky edges are blurred or over-smoothed in the motion-compensated prediction. In this thesis, we propose an improved OBMC scheme that adopts adpative weighting matrices for the case of variable block size motion compensation. Specifically, we resort to a parametric solution in closed form to express the optimal weights for the OBMC windows, based on the spatially-variant prediction of motion vectors. Experimental results show significant improvement over existing successful 3D wavelet-based scalable video codecs in both objective and subjective quality. Besides, the average encoding time can be saved by 14.81%. KEY WORDS; Video coding; 3-D Wavelet transform; Rate distortion optimization; Lagrange multiplier; Overlapped block motion compensation TYPE OF DISSERTATION: Application Fundamentals