基于GPU的视频转码技术研究

视频转码是个复杂的过程,它需要对已经压缩过的码流进行解析,然后经过处理转换成满足解码终端要求的目标格式码流。为了提高视频转码的效率并降低视频转码的计算复杂度,根据视频转码的要求和图形处理器的并行结构,提出了一种利用GPU强大的并行计算能力来加速视频转码的算法。该算法将视频转码过程中耗时最多、最复杂的运动估计和模式选择过程转移到GPU上并行执行。在开发GPU通用计算能力的时候,采用NVIDIA公司的CUDA(统一计算设备架构)计算平台。实验结果证明,该算法可以有效提高视频转码的速度和效率。     

面向在线制作的集群转码系统设计

根据广域网环境的在线节目制作系统的需求,设计了一种集群转码系统。该系统采用基于Condor的集群运算调度机制,结合对视频数据的有效分割,完成多运算节点同时对一个节目素材的转码计算。重点介绍了集群转码系统的总体设计思路以及实现方式,包括软件部分任务生成、转码程序、结果生成等模块的实现细节。系统测试数据说明了集群系统中视频分割大小以及集群节点数目等关键因素对系统运行效率的影响。     

基于超算平台和Hadoop的并行转码方案设计

随着三网的快速发展和融合,海量视频转码需求日益增长。在分析了目前主流的海量视频转码方案的基础上,提出一种基于超算平台和Hadoop的并行转码方案。在方案中,利用超算平台的计算节点搭建一个基于Hadoop的并行转码系统,在MapReduce框架下完成对视频的分割、并行转码与合并。同时,结合超算平台的高运算能力、高主存容量、多核多线程等特点和MIC架构的强浮点运算能力和高度并行能力,提出特殊的技术实现策略,达到加速转码的目的。     

Research on Scheduling Algorithms for Distributed Video Transcoding Service

针对分布式环境下视频转码服务的任务请求自适应接收问题以及处理后视频序列分段向客户端的端传输控制问题,通过分析视频转码处理对资源占用的特点以及流式媒体服务实现的过程,提出了基于处理节点资源与客户信道质量反馈的视频转码任务自适应接收策略和用于实现客户端流畅视频服务的视频任务分段传输控制算法.实验表明,相比现有算法,该自适应客户请求接入策略能够依据客户端的信道质量选择最适合的任务版本;在用户播放缓冲区有限的条件下,提出的传输控制算法可以达到更小的任务失败几率.     

分布式视频转码服务调度算法研究

针对分布式环境下视频转码服务的任务请求自适应接收问题以及处理后视频序列分段向客户端的端传输控制问题,通过分析视频转码处理对资源占用的特点以及流式媒体服务实现的过程,提出了基于处理节点资源与客户信道质量反馈的视频转码任务自适应接收策略和用于实现客户端流畅视频服务的视频任务分段传输控制算法。实验表明,相比现有算法,该自适应客户请求接入策略能够依据客户端的信道质量选择最适合的任务版本;在用户播放缓冲区有限的条件下,提出的传输控制算法可以达到更小的任务失败几率。     

MPEG-2与MPEG-4间转码的关键技术研究与实现

视频转码技术的出现，解决了视频通信领域数据在不同系统和设备间互通互连的问题。本文主要研究了MPEG-2到MPEG-4的转换编码，提出了运动矢量映射的转码结构，并对其中的关键算法进行了研究和实现。仿真试验表明，本文结构和算法在保证了转换编码灵活性的同时，具有较好的转码效率和鲁棒性。     

DCT域视频转码技术综述

随着数字视频技术的广泛应用,实现各种视频编码格式之间转码的要求越来越迫.视频转码是数字媒体网络网关的关键技术.本文从DCT域视频转码的体系结构、关键技术及最优转码策略三个方面,介绍了现有的各种算法,分析和总结了各自的特点,提出了DCT域视频下采样帧内刷新体系结构和最优视频转码策略解的概念,并指出了下一步的研究方向.     

一种新的MPEG-2到H.264的转码算法

研究了MPEG-2到H.264的转码技术,在MPEG-2和H.264压缩算法基础上提出了一种基于MV重用和H.264多参考帧特性的转码算法.实验表明,本算法与反向查找表算法相比在转码效率和转码视频质量两方面都有所提高.     

低复杂度的快速降尺寸视频转码算法

为了降低降尺寸视频转码的运算复杂度,提出一种基于支持向量机的快速转码模式决策算法。首先从输入的高分辨率视频码流中的编码信息里选取多维特征向量,并选择与模式特征匹配的核函数训练SVM分类器模型,建立高分辨率视频编码信息与降尺寸视频宏块编码模式之间的相关性;然后构建分层式SVM分类器对降尺寸视频中宏块模式进行阶梯式预测分类,以此缩减预测模式数量,提高转码效率。实验结果证明,算法可以节省高达67.31%的运算量,同时保证转码后视频的高质量。     

一种用于H.264视频容错转码的帧内宏块刷新算法

无线网络信道具有误码率高、延时抖动等不稳定的特性。为了保证视频信息从有线网络到无线网络中传输的质量,需要对视频进行容错转码。基于H.264容错转码系统,利用转码结构中解码部分的运动矢量信息,自适应地在编码部分进行帧内刷新,以提高视频的抗误码能力。实验结果对比表明,该帧内刷新算法能够以较少的比特率为代价,有效地提高了接收端的视频质量。     

Framework for Fast and Efficient Cloud Video Transcoding System Using Intelligent Splitter and Hadoop MapReduce

The way computing was done has changed a lot in recent times. Nowadays mobile devices have been the stealing the show. These devices come in different specifications, that any multimedia content that is to be played requires transcoding for better user experiences. Cloud based video services cater to the needs of the end user based on their requirements, through video transcoding. Hence video transcoding plays a very important role in today’s evolving streaming media environment. The major problem with video transcoding is that it consumes a lot of time and impacts seriously on the quality of the output. Transcoding uses the device information to transform the video into the required format and this process is done in a distributed fashion, to speed up the process. This work proposes an Intelligent Video Splitter which uses the Map Reduce algorithm to provide efficiency based on time factor. The important performance metrics including video distortion (VD), video distortion due to frame dependency (FDD) were considered. The results showed that the proposed framework perceptibly outperforms than the prevailing strategies. It provides higher video quality as a result of it introduces less video distortion. In future this method may be extended to supply associate automatic device aware video standards.     

MPEG-2到H.264转码中变换系数转换的优化算法

为采用优势明显的H.264编码,并避免已有的大量MPEG-2非线性视频编缉设备被淘汰,研究了从MPEG-2到H.264转码的转换技术,分析了目前的转换算法,提出了一种DCT系数到H.264整数变换(HT)系数转换的优化算法,并针对不同视频终端的应用需求,分别给出了保持图像空间分辨率不变和减半时的转换方法.实验表明,该算法可大幅度简化运算,并具有良好的实时性和视频质量.     

DVC转码技术研究

本文介绍了DVC(分布式视频编码)到传统视频转码的方案,该转码方案适用于移动终端设备之间的视频通信。着重讲述了DVC到H.264的转码,针对转码过程中复杂度高和时延长等问题,利用DVC解码端生成的运动矢量来减少H.264编码的工作量,在几乎不影响视频质量的前提下,极大地降低了转码的计算复杂度和时间,提高了转码效率。同时介绍了DVC转到其它传统视频的方法和方案,最后分析了DVC转码在当前移动视频通信市场中存在的巨大潜能,以及对转码技术的未来发展和方向进行了展望。     

On the methods and performances of rational downsizing video transcoding

Video transcoding is a popular technique for adapting the bit-rate or spatial/temporal resolution of a precoded video to suit better the constraints and requirements of different transmission networks and receiving devices. To minimize computational complexity, many fast methods have been proposed to obtain the motion vectors required for transcoding a precoded video through reducing its frame size by an integral factor. In this paper, we extend the existing work by developing and comparing several fast methods of downsizing precoded videos by a rational factor. Methods that outperform others under different conditions or with different computational requirements are identified, and an application scenario that can benefit from the proposed rational downsizing video transcoding is presented. An efficient scheme is also proposed to select the proper reduced frame size for sustaining the best possible video quality at a specified lower bit-rate. The superiority of the proposed transcoding approach in comparison with the existing integral downsizing video transcoding or cascaded video re-encoding methods is evident from the experimental results shown in this paper.     

PPmcast: a novel P2P-based application-level multicast using Kamulia

The rapid growth and popularity of networked meeting environments such as video conference in recent years have spawned a series of research interests in constructing large-scale environments. For increasing scalability and decreasing the cost of management and deployment, more and more studies propose using peer-to-peer (P2P) architectures to construct large-scale application-level multicast algorithm for games, multimedia and other applications. In order to improve the applications' performance, high efficiency multicast algorithm is required. This paper focuses on developing a novel P2P application-level multicast algorithm based on Kamulia protocol. In the proposed algorithm, all nodes of P2P network are organized into structured overlay network by distributed hash table (DHT). The distance metric between two nodes is obtained by computing the exclusive or (XOR) value. The theoretical analysis indicates the proposed algorithm has better time complexity degree compared with other similar algorithms. The simulation experiments also show that the proposed algorithm has better performance on the following aspects, control expenses, loss ratio after failure, and waiting time to the first packet.     

针对移动视频终端的基于感兴趣区域的快速转换编码

提出一种针对移动终端,基于感兴趣区域(ROI)的快速转换编码方案.首先,根据移动终端的显示尺寸,在视频服务器端利用视觉关注度模型从H.264视频流自动地检测出ROI.然后,在代理服务器端根据ROI转换编码生成适合于移动终端的视频流.此外,针对此转码体系提出了一种快速模式选择算法.仿真实验结果表明,本方案可在降低网络占用带宽的情况下,获得较好的主观视觉效果,并且计算量小.     

Foveation embedded DCT domain video transcoding

Video transcoding is a key technology to support video communications over heterogeneous networks. Although quite a bit of research effort has been made in video transcoding due to its wide applications, most video transcoding techniques proposed in the literature are optimized based on the simple mean squared error (MSE) metric which does not correlate well with the human visual perception. In this paper, foveation, a property of the HVS, is exploited in video transcoding. The proposed foveation embedded DCT domain video transcoding can reduce the bit rate without compromising visual quality or achieve better subjective quality for a given bit rate by shaping the compression distortion according to the foveated contrast sensitivity function of the HVS. In addition, fast algorithms for video foveation filtering and DCT domain inverse motion compensation are developed, which significantly improve the efficiency of video transcoding.