【摘要】 针对IP网络(如因特网)带宽随时变化的特性,传输视频时,为了有效地利用现有带宽,尽可能提高视觉质量,该文在MPEG-4精细可伸缩性视频编码(FGS)的选择性增强技术基础上提出一种基于运动矢量的自适应选择性增强算法。通过提取、分析基本层运动矢量,自适应地确定目标区域,并将结果在增强层编码用于选择性增强。使FGS编码实现基于内容的增强和优先传输感兴趣区域(目标区域)。实验表明该算法使目标区域在不同的码率下都能获得相对最佳的视频质量,适合网上视频监控之类背景静止而人们仅对场景中运动的人或物感兴趣的应用。更多还原

【Abstract】 Video transmissions over IP networks (i.e. Internet) encounter variable bandwidth transmissions. To efficiently utilize the given bandwith to obtain the best possible visual quality, an adaptive selective enhancement algorithm employing motion vectors was developed based on the selective enhancement mechanism of the FineGranularityScalability (FGS) coding scheme of MPEG4. Motion vectors generated during baselayer encoding were analyzed to adaptively locate object regions. The result was then used for selective enhancement during enhancement layer encoding. The proposed algorithm enables FGS to perform contentbased enhancement and prioritized transmission of important regions (object regions). The algorithm significantly improved the video quality of important regions at different bitrates. The algorthm is suitable for applications such as internet surveillance where the backgrounds are still and only the moving objects within the scene are of interest. 更多还原