基于运动区域检测的会议视频清晰度增强处理  

Enhancement of Conference Video Definition Based on Motion Region Detection

作　　者：彭程 PENG Cheng(China Mobile Communication Group Guangxi Co.,Ltd.,Nanning 450000,China)

机构地区：[1]中国移动通信集团广西有限公司,广西南宁450000

出　　处：《电视技术》2021年第3期18-20,共3页Video Engineering

摘　　要：视频会议场景对视频增强的实时性有较高的要求。针对现有视频增强算法(如BM3D等)存在的耗时长的问题,基于变化区域检测提出一种基于运动区域检测的视频增强技术。对时序帧数据进行颜色空间转换,快速地将视频场景分为静止区域和运动区域,之后对静止区域进行时域降噪和视频增强。在视频帧集合的实验结果表明,该算法可以显著地增强图像纹理细节,并具有很强的时效性,在视频会议场景有很大的应用潜力。Video conference scene has a high demand for real-time video enhancement.Aiming at the time-consuming problem of existing video enhancement algorithms,such as BM3D,a video enhancement technology based on motion region detection is proposed.The temporal frame data is converted into color space,and the video scene is quickly divided into static region and moving region.Then the static region is denoised and enhanced in time domain.The experimental results in video frame set show that the algorithm can significantly enhance the image texture details,and has a strong timeliness,which has great application potential in video conference scene.

关 键 词：视频增强 运动检测 时域降噪 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]