浅析背景替换技术在视频会议中的应用  

Analysis on the Background Replacement Technology in Video Conference

作　　者：邵晓文 SHAO Xiaowen(Renmin University of China,Beijing 100000)

机构地区：[1]中国人民大学,北京100000

出　　处：《软件》2022年第1期108-111,114,共5页Software

摘　　要：如今,随着网络带宽的飞速增长,越来越多的人会倾向于使用视频通话的方式进行沟通。而随着全球新冠疫情的蔓延,为了保持社交距离,许多公司或团体开始依赖使用视频会议的方式开展工作。背景替换技术可以在视频会议的过程中,保护隐私和提高观众的专注度。本文将对两款基于深度学习的图像处理框架,应用在背景替换时的情况,进行对比和分析,并给出了针对于不同场景下如何选择的建议。Nowadays, more and more people tend to use video calls to communicate since bandwidth of the internet has grown rapidly. To keep social distance, many companies or groups have begun to rely on video conference to carry out their daily job because the COVID-19 pandemic spread quickly. The background replacement technology can protect the privacy and keep the concentration of the audience during video meeting. The article will focus on two types of image processing frameworks based on deep learning. After comparing and analyzing, we will give the proposal on how to choose these two frameworks in various scenarios.

关 键 词：背景替换 视频会议 背景虚化 深度学习 动态背景 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]