一种利用视觉掩蔽效应的感兴趣区域视频编码方法

传统感兴趣区域视频编码方法通过整体降低ROI区域的量化参数来实现ROI质量优化,会造成整体率失真性能的下降,尤其在低码率时下降更加严重。针对这一问题,结合人眼掩蔽特性的精细Grid量化方法在文中提出,可有效提高ROI质量并降低整体质量的损失。而且该量化方法能够根据ROI尺寸大小对初始感知权重进行自适应调节,适用于ROI特征不同的视频序列。实验结果表明,对于不同特征的标准测试序列,在中低码率条件下和H.264相比,本文算法能平均提高ROI的PSNR值1.905dB,主观感受有明显提升。和同类ROI算法相比,ROI质量平均提升0.47dB,整体质量提高0.625dB。     

一种基于感兴趣区域的图像检索算法

提出了一种基于感兴趣区域的图像检索算法，该算法以图像中用户感兴趣的目标为研究对象，使用一种改进了的区域分割算法对图像进行目标区域分割，采用伪Zemike矩提取分割出来的区域形状特征结合其它区域特征构成检索特征向量。本文还讨论了在检索系统中用户感兴趣的区域与分割后图像目标区域的映射关系。在实验系统中证明有较高的区域准确率。     

一种基于优先级的多个感兴趣区域图像编码方法

JPEG2000采用感兴趣区域编码能够实现一幅图像的多种质量压缩。针对以往提供的各种方法都对一个感兴趣区域进行编码的缺点,提出了实现多个感兴趣区域的无损压缩的方法。其主要特点是利用优先级的方法重新排列编码数据流中的包,通过优先级排列可以区分ROI和背景区域在每个分解级中的相对重要性,而且在同一幅图像中有多个感兴趣区域的时候,它能够提供多个优先级,能够很好的实现多个感兴趣区域无损压缩的图像编码。     

一种新的感兴趣区域编码方法

感兴趣区域(ROI)编码和渐进解码都是JPEG2000的重要特性。当前的ROI编码算法不能控制渐进解码图像的相对质量或是计算量太大。基于位面组移动法和分段的思想,该文提出了位面群移动法,使位面的排列更加灵活,不同码率的解码图像相对质量更加稳定。该方法仅增加少量数据开销,计算复杂度也没有明显增加,实现比较简单。     

一种基于用户感兴趣区域的图像检索算法

利用图像全局特征的检索不能很好地检索用户想要的对象内容,而基于分割后各区域特征的检索又过分依赖于复杂的图像分割算法。针对上述两者的缺点,文中提出了一种基于用户感兴趣区域的图像检索算法。该算法首先对样例进行多分辨率树状分解,再由用户选择分解后的任意多个感兴趣的子图,提取子图的特征以进行相似性度量,并应用相关反馈以更好地捕获用户的检索意图。该方法无需对图像进行复杂的分割就能提取对象特征,且经由实验证明具有较高的查全率。     

基于交互式感兴趣区域的教学视频编码

为了在现有网络带宽和网络客户端处理能力的条件下提高网络多媒体教学系统中的视频质量, 提出一种基于交互式感兴趣区域的教学视频编码算法。首先利用经典的GrabCut算法在低分辨率视频中交互式检测感兴趣目标, 基于区域生长算法定位各目标所在区域并映射到高分辨率视频中; 然后提取出感兴趣区域, 缩放到显示分辨率, 对各感兴趣区域编码生成H. 264标准兼容的码流。实验结果表明, 在相同视频质量下, 该算法可以有效减小传输码率、降低客户端的计算复杂度, 相比现存方法, 传输码率最大节约94. 3%, 计算复杂度降低526%。     

一种基于感兴趣区域提取的新型SVD图像压缩算法

提出了一种基于感兴趣区域提取的新型奇异值分解（SVD）图像编码算法（SODSVD）。根据图像的区域复杂度、视觉注意等指标运用感兴趣区域提取技术将图像自适应分块,而对每一小块,根据其复杂度进行不同秩的奇异值分解算法,从而达到图像压缩的目的。同时还提出了一种简单的基于感兴趣区域的图像划分策略。实验结果表明,在PSNR和bpp等评价指标上,新方法有明显优势。     

一种基于用户感兴趣区域的图像检索算法

利用图像全局特征的检索不能很好地检索用户想要的对象内容，而基于分割后各区域特征的检索又过分依赖于复杂的图像分割算法。针对上述两者的缺点，文中提出了一种基于用户感兴趣区域的图像检索算法。该算法首先对样例进行多分辨率树状分解，再由用户选择分解后的任意多个感兴趣的子图，提取子图的特征以进行相似性度量，并应用相关反馈以更好地捕获用户的检索意图。该方法无需对图像进行复杂的分割就能提取对象特征，且经由实验证明具有较高的查全率。     

面向室外视频监视的感兴趣区域提取

针对室外视频监视中运动对象检测易受树枝叶晃动、水面波动等无意义运动干扰,准确性低、实时性差的问题,定义感兴趣区域为已经存在及潜在存在有意义运动对象的区域,提出一种感兴趣区域自动提取算法。构造带状算子提取训练阶段存在有意义运动对象的区域,利用颜色一致区域生长和干扰对象区域退化得到潜在存在有意义运动对象的区域,对不同区域采取不同的检测策略可以提高检测的准确性和实时性。实验结果表明,该算法对感兴趣区域提取结果良好,用于室外视频监视中运动对象的检测能克服无意义运动干扰,提高检测的准确性,并能有效减少计算量。     

一种遥感图像中多感兴趣区域的压缩算法

遥感图像的特殊应用,常要求对遥感图像进行无损压缩。但是由于遥感图像含有丰富的纹理,获得的无损压缩比一直很低。而遥感应用中,常常是对某一地区和某几个地区感兴趣,所以可以对感兴趣区域进行无损压缩,而以损失非感兴趣区域信息为代价。本文提出了一种用于遥感图像的多感兴趣区域的压缩算法LS-MROI,实验结果表明该算法用于ROI区域压缩能够获得较好的压缩性能。     

基于感兴趣区域的H.264视频加密算法

为了解决现有的基于H.264的加密算法无法满足安全性和加密效率之间较好折衷的问题,提出两种基于感兴趣区域的H.264加密算法,将感兴趣区域的提取和基于熵编码的H.264加密算法相结合,只对提取出来的感兴趣区域进行加密.设计了一种基于人脸检测的加密算法,并通过修改模式选择算法,去掉了使得非人脸区域参考人脸区域的帧间宏块预测类型,解决了由帧间预测引起的人脸加密区域变形的问题.将运动对象的检测与提取和H.264的编码过程结合,设计了一种基于运动对象检测的H.264加密算法.实验结果表明,这两种加密算法能够在不降低算法安全性的情况下获得安全性和加密效率之间的较好折衷,可以满足实时应用的需求.     

H.264到AVS快速视频转码技术研究

通过深入对比和分析H.264 Main Profile和AVS-P2的L4.2两种视频编解码标准,本文提出一种快速实现H.264到AVS视频转码的技术,并对该转码技术进行软件实现。     

OpTrans:一个率失真优化的MC-EZBC转换编码算法

可伸缩性视频转换编码算法是大型分层视频组播系统层次化代理自适应机制的核心技术.好的转换编码算法不仅要满足拥塞速率控制的要求,而且要使分层视频组播流中的增强层数据以率失真优化的方式进行截断,以提高接收者的接收效能.本文对可伸缩三维小波编码算法MC-EZBC视频转换编码的联合率失真目标函数和优化算法进行了深入研究,提出了率失真优化的MC-EZBC转换编码算法及其分层速率控制方法.对比实验结果表明,采用了联合率失真优化后的转换编码码率控制方法比原有的按顺序截断方法可提高平均视频质量PSNR0.3-0.5dB,进而可优化大型分层视频组播系统的整体效能.     

一种基于分层率失真优化的容错性视频转码算法

容错性视频转码能够在已压缩视频流中嵌入容错性工具,以增强视频流面向无线信道的抗误码能力.但是容错性工具的嵌入在降低失真的同时也会增加码率,因此需要进行率失真优化.针对这一问题,文中提出了一种基于分层率失真优化的容错性转码算法.该算法通过在帧层和宏块层分别对重同步标记的插人和帧内宏块的刷新进行不同粒度的优化来提高算法的灵活性,通过在帧层考虑帧内宏块刷新、帧内预测和运动矢量预测的影响来提高算法的精确性,通过在宏块层减少候选模式的数量来降低算法的复杂度.实验表明,该算法与基于单层率失真优化的算法相比,信噪比可获得0.6～1.1dB的增益,复杂度最多可降低25%.     

恒定速率视频编码转换码率控制策略

分析了恒定速率视频编码转换缓冲区的状态，导出了编码转换缓冲区为防止解码器缓冲区下溢和上溢应满足的条件，建立了视频编码转换模型。并使用小波变换系数来表征图像特征，提出了基于图像特征的码率控制策略。模拟实验表明，该码率控制策略能有效地减少、避免缓冲器出现上、下溢的情况，使输出码率趋于稳定，提高了重建视频图像的信噪比。     

降低空间分辨率视频编码转换方法

根据漂移误差与帧间编码块的运动活动性的关系,提出了一种自适应帧内刷新方法。该方法根据目标比特率和实际比特率的差值,动态地调整阈值,从而有效地减少了帧内编码的比特率,保持了对差错的鲁棒性,限制了差错的时间传播。同时,根据率失真定理,在保证失真最小的情况下,为帧内不同的宏块选用最佳的量化系数。仿真结果表明:文中提出的方法有效地平滑了缓冲器的输出,同时峰值信噪比也有所提高。     

基于CIF-to-QCIF变换的压缩域视频水印算法

现有视频水印算法的设计目标主要是对有损压缩、重量化编码等处理具有鲁棒性,但很少考虑视频帧格式变换因素,使水印对这类处理不具有鲁棒性。论文结合视频帧格式由CIF到QCIF的变换设计了一个压缩域的视频盲水印算法,水印被嵌入到CIF格式的视频流中,既可以从CIF视频中提取也可以从QCIF视频中提取,实验结果表明提出的算法对MPEG重编码以及CIF到QCIF的视频格式变换具有鲁棒性,而且整个嵌入提取过程都基于DCT域,具有较低的计算复杂度。     

A QoS-Enabled Transmission Scheme for MPEG Video Streaming

While MPEG is the de facto encoding standard for video services, online video streaming service is becoming popular over the open network such as the Internet. As the performance of open network is non-predictable and uncontrollable, the tuning of the quality of service (QoS) for on-line video streaming services is difficult. In order to provide better QoS for the delivery of videos, there are proposals of new encoding formats or new transmission protocols for on-line video streaming. However, these results are not compatible with popular video players or network protocols and hence these approaches are so far not very successful. We use another approach which tries to by-pass these problems. We designed a QoS Tuning Scheme and a QoS-Enabled Transmission Scheme for transmitting MPEG videos from video servers to clients. According to the traffic characteristics between the video server and each individual client, the QoS Tuning Scheme tunes the QoS to be delivered to each individual client on the fly. Furthermore, our QoS-Enabled Transmission Scheme can be applied over any protocol, such as HTTP which is the most popular protocol over the open network. With our transmission scheme, bandwidth can be better utilized by reducing transmitted frames which would have missed their deadlines and would eventually be discarded by the clients. This is achieved by sending frames according to their impact on the QoS in the playback under the allowed throughput. With these schemes, users can enjoy video streaming through their favorite video players and with the best possible QoS. In order to facilitate the real time QoS tuning, a metric, QoS-GFS, is developed. This QoS-GFS is extended from the QoS-Index, another metric which has taken human perspective in the measurement of video quality. Hence QoS-GFS is better than the common metrics which measures QoS by means of rate of transmission of bytes or MPEG frames. We designed and implemented a middleware to perform empirical tests of the proposed transmission scheme and QoS tuning scheme. Experiment results show that our schemes can effectively enhance the QoS for online MPEG video streaming services. The work reported in this paper was supported in part by the RGC Earmarked Research Grant under RGC HKBU 2074/01E, and by the FRG under FRG 00-01/I. Joseph Kee-Yin Ng received a B.Sc. in Mathematics and Computer Science, a M.Sc. in Computer Science, and a Ph.D. in Computer Science from the University of Illinois at Urbana-Champaign in the years 1986, 1988, and 1993, respectively. Dr. Ng is currently an associate professor in the Department of Computer Science at Hong Kong Baptist University. His current research interests includes Real-Time Networks, Multimedia Communications, Ubiquitous/Pervasive Computing, Mobile and Location-aware Computing, Performance Evaluation, Parallel and Distributed Computing. Dr. Ng is the Technical Program Chair for TENCON 2006, General Chair for The 11th International Conference on Embedded and Real-Time Computing Systems and Applications (RTCSA 2005), Program Vice Chair for The 11th International Conference on Parallel and Distributed Systems (ICPADS 2005), Program Area-Chair for The 18th & 19th International Conference on Advanced Information Networking and Applications (AINA 2004 & AINA 2005) and he had served as the General Co-Chair for The International Computer Congress 1999 & 2001 (ICC'99 & ICC'01), the Program Co-Chair for The Sixth International Conference on Real-Time Computing Systems and Applications (RTCSA'99) and the General Co-Chair for The 1999 and 2001 International Computer Science Conference (ICSC'99 & ICSC'01). Dr. Ng is a member of the Editorial Board of Journal of Pervasive Computing and Communications, Associate Editor of Real-Time Systems Journal and Journal of Mobile Multimedia. He is a guest editor of International Journal of Wireless and Mobile Computing for a special issue on Applications, Services, and Infrastructures for Wireless and Mobile Computing. Dr. Ng is currently the Region 10 Coordinator for the Chapter Activities Board of the IEEE Computer Society, and is the Coordinator of the IEEE Computer Society Distinguished Visitors Program (Asia/Pacific). He is a senior member of the IEEE and has been a member of the IEEE Computer Society since 1991. Dr. Ng has been an exco-member (1993–95), General Secretary (1995–1997), Vice-Chair (1997–1999), Chair (1999–2001) and is the immediate past Chairman of the IEEE, Hong Kong Section, Computer Chapter. Dr. Ng received the Certificate of Appreciation for Leadership and Service (2000–2001) from IEEE Region 10 and the IEEE Meritorious Service Award from IEEE Computer Society at 2004. He is also a member of the IEEE Communication Society, and ACM. Karl R.P.H. Leung received his Ph.D. from The University of Hong Kong. He is currently a Principal Lecturer in the Department of Information and Communications Technology at the Hong Kong Institute of Vocational Education (IVE). He is the founder of the Compuware Software Testing Laboratory in the IVE with a donation from the Compuware Asia Pacific Co. Ltd. His research areas include: domain modeling, mission critical software engineering methodology, secure workflow systems, GSM-based location estimation, and QoS of MPEG streaming. He is a Senior Member of the IEEE and IEEE Computer Society, and has held major office of the IEEE Hong Kong Section Computer Chapter. While he was the chairman in 1998, the Chapter won the IEEE Most Outstanding Computer Society Chapter Award. He is also a Chartered Engineer of Engineering Council (UK), a Chartered Information Systems Engineer of British Computer Society (UK), an Engineer of Hong Kong Institution of Engineers, Registered Professional Engineer (Information) of Hong Kong Engineers Registration Board, and a member of ACM, BCS, ACS, HKIE and HKCS. Calvin Kin Cheung Hui received a B.Sc. (First Class Honours) in Computer Science, and a M.Phil. degree in Computer Science from Hong Kong Baptist University in the years 1999, and 2002, respectively. Mr. Hui's research interests includes Real-Time Networks, VoD Systems, Video Streaming, Multimedia Communication, and Distributed Systems Performance Evaluation.     

基于视点跟踪的感兴趣区检测方法

针对目前基于内容的图像检索系统中,难以客观弥补语义鸿沟的问题,提出了一种利用视点跟踪技术检测感兴趣区的方法.人眼注视运动传递了大量反映个体心理活动的信息.通过实时捕获人眼的注视点,客观地获得用户兴趣信息,根据注视点位置计算用户兴趣度,最终提取图像感兴趣区.本文提出了一种低复杂度的实时视点跟踪方法,定义了兴趣度的度量公式,并建立了一套基于视点跟踪的感兴趣区检测实验平台.实验结果表明,该方法可以有效地提取用户感兴趣区,弥补语义鸿沟.