Novel video coding, transcoding and multicoding technologies applied to H.264 and HEVC

在未来几年内，无线和有线视听服务有望融合，与此同时，移动的视频和IPTV用户数量将出现惊人增长。这些服务将成为其运营商和提供商的重要收入来源，视频编码和转码将在这些服务和其他服务中发挥关键作用，使他们能够与异构终端和网络一起工作。然而，视频的日益普及给我们的信息技术（IT）基础设施带来了越来越大的压力，包括计算复杂性、带宽和存储要求。该项目将通过研究在两个视频研究领域应用于H.264和HEVC的新技术来解决这些问题。在第一个研究领域，我们将致力于提高视频的视觉质量：基于新颖的低复杂度高精度视觉质量指标的感知失真率优化方法，使用机器学习的近最佳帧级视频编码参数估计，以及一种速率控制算法（基于最近的失真模型），提供一致的帧到帧质量。在第二个领域，我们将致力于降低视频编码和转码操作的计算复杂性：高效的视频转码，利用源的运动矢量，编码模式和残留信息，用于H.264到HEVC转码，基于运动场和率失真（R-D）模型的新型多流视频编码器/转码器，以及大规模并行处理器上的HEVC编码。解决这些问题将有助于下一代视频应用和服务的顺利和经济高效的部署。 这对加拿大电信业（特别是移动的运营商和IPTV服务提供商，如有线电视运营商）非常重要，对他们下一代多媒体应用和服务的成功至关重要。最后，它将有助于加强加拿大公司Vanessa在转码领域的技术领导地位，Vanessa是视频优化和交付解决方案的全球领导者。

Within the next few years, a convergence of wireless and wired audio-visual services is expected, and, at the same time, phenomenal growth in the number of mobile video and IPTV users. These services will represent an important source of revenue for their operators and providers, and video coding and transcoding will play a key role in these and other services to enable them to function with heterogeneous terminals and networks. However, the growing popularity of video is putting heavy and growing pressure on our information technology (IT) infrastructures, in terms of computational complexity, bandwidth and storage requirements. This project will address these problems by investigating novel technologies applied to H.264 and HEVC in two video research areas. In the first research area, we will work towards improving the visual quality of videos: perceptual distortion rate optimization methods based on novel low-complexity high-accuracy visual quality metrics, near-optimal frame level video coding parameter estimation using machine learning, and a rate control algorithm (based on recent distortion models) giving consistent frame-to-frame quality. In the second area, we will work towards reducing the computational complexity of video coding and transcoding operations: efficient video transcoding exploiting the source's motion vectors, coding modes and residual information for H.264 to HEVC transcoding, novel multi-stream video encoder/transcoder based on motion fields and rate-distortion (R-D) models, and HEVC encoding on massively parallel processors. Solving these problems will contribute to a smooth and cost-effective deployment of next generation video applications and services. This is very important to the Canadian telecommunications industry (especially mobile operators and IPTV service providers, such as cable operators) and crucial to the success of their next generation multimedia applications and services. Finally, it will contribute to strengthening the technological leadership in the transcoding field of Vantrix, a Canadian company and world leader in video optimization and delivery solutions.