计算机技术与发展2016，Vol.26Issue(10)：87-91,5.DOI:10.3969/j.issn.1673-629X.2016.10.019
基于台标识别的网络视频分类系统
A Classification System for Internet Videos Based on TV Logo Recognition
摘要
Abstract
In order to clean up the illegal videos on the internet rapidly and accurately and reduce the input of manpower and material re-sources,a video classification system based on TV logo recognition has been proposed. On the basis of analyzing illegal videos TV logo with different producer,the system adopts a method combined pattern matching and feature recognition for TV logo recognition to im-prove accuracy according to different TV logo feature of illegal video. At the same time,a variety of image enhancement techniques are a-dopted to decrease the influence of uncertain factors like resolution,definition and background on the identification results,improving the recall rate of illegal video and recall accuracy. Experimental results show that this system could effectively reduce the human input and material resource by screening out the security video and suspected illegal video. In addition,FFMPEG is used to decode and draw key frame,which can improve the efficiency at 30%.关键词
公共安全/网络视频/台标识别/模板匹配/ORB特征识别Key words
public safety/internet videos/TV logo recognition/template matching/ORB分类
信息技术与安全科学引用本文复制引用
刘缨,尹珩,戴文娟..基于台标识别的网络视频分类系统[J].计算机技术与发展,2016,26(10):87-91,5.基金项目
北京市科委项目(Z141100006014002) （Z141100006014002）
