摘要
互联网流媒体平台上的个人直播模式正在盛行,成为人们日常交流、消闲的主要手段。本文提出了一种用来监控实时直播流,并根据感知状态和帧差分析智能识别非法的实时不当行为。所提出的系统利用聊天室状态的若干指标而不是单独和直接地分析图像;通过监控某实时直播平台验证监控效率,三个指标的整体准确率达到90.1%,该系统的应用可以改变现场平台监督完全依赖于人工审查的现状。
The personal live broadcast mode on the Internet streaming media platform is prevailing and becomes the main means of daily communication and leisure.This paper proposes a method for monitoring real-time live broadcasts and intelligently identifying illegal real-time misbehavior based on perceived state and frame difference analysis.The proposed system utilizes several indicators of chat room status instead of analyzing images directly and directly;by monitoring a real-time live broadcast platform to verify monitoring efficiency,the overall accuracy of the three indicators reaches 90.1%,and the application of the system can change the on-site platform supervision.It is completely dependent on the status quo of manual review.
作者
杜雨潇
DU Yu-xiao(Xi'an Aeronautical Polytechnic Institute·School of Aeronautical Management Engineering,Xi'an 710089 China)
出处
《自动化技术与应用》
2020年第10期159-162,共4页
Techniques of Automation and Applications
基金
西安航空职业技术学院教改课题(编号18XHJG-022)
西安航空职业技术学院科研课题(编号18XHGZ-09)
西安航空职业技术学院科研课题(编号18XHGZ-012)。
关键词
互联网流媒体
内容监督
交互式直播
图像分析
Internet streaming
content supervision
interactive live broadcast
image analysis
