摘要
以路况实时监控视频为例,展示了将SVC应用于WebRTC-SVC中时的性能。将信息分在基础层和增强层两个通道来进行传输,在基础层中传输语义信息,而在增强层中传输背景信息。此外提出一种基于SMDC的高效压缩方法,通过语义运动分析来识别关键帧实现压缩,同时设计插值机制弥补流畅性损失。通过WebRTC-SVC和SMDC算法大大降低了机器视觉识别场景下的带宽需求。端到端测试证明,相比H.264,此方法在WebRTC场景下更高效,即使在较低比特率下也能传输高质量视频。
Taking real-time traffic monitoring video as an example,the scalable video coding(SVC)performance is demonstrated with its application in WebRTC.By segmenting the information into the base layer and the enhancement layer channels,semantic information is transmitted in the former,while background information is conveyed in the latter,Furthermore,an efficient compression method based on semantic motion detection and compression(SMDC)is introduced,utilizing the semantic motion analysis to identify key frames for compression and addressing the fluency loss issue with an interpolation mechanism.The integration of WebRTC-SVC and the SMDC algorithm significantly reduces the bandwidth requirements for machine vision recognition scenarios.The end-to-end testing validates that this approach outperforms H.264 and can deliver high-quality videos even at lower bit rates in WebRTC scenarios.
作者
牛冠冲
刘飞翔
杨雯
曹琦
白宝明
NIU Guanchong;LIU Feixiang;YANG Wen;CAO Qi;BAI Baoming(Guangzhou Institute of Technology,Xidian University,Guangzhou 510555,China;School of Telecommunications Engineering,Xidian University,Xi'an 710071,China)
出处
《移动通信》
2024年第2期47-55,共9页
Mobile Communications
基金
中央高校基本科研业务费专项资金“智能协同系统的通感一体化理论研究”(20103237888)
中央高校基本科研业务费专项资金“基于元宇宙的教学平台建设”(20199236467)
国家自然科学基金“基于语义通信的零错误传输”(62301406)
国家重点研发项目“面向未来无线通信信息处理若干关键问题的数学理论和方法”(2021YFA1000500)。
