项目名称: 面向内容安全监控应用的视频指纹实时获取与快速检索技术研究
项目编号: No.61201128
项目类型: 青年科学基金项目
立项/批准年度: 2013
项目学科: 电子学与信息系统
项目作者: 徐杰
作者单位: 电子科技大学
项目金额: 25万元
中文摘要: 在当前网络中,视频业务已成为网络的主体业务,然而视频在传播过程中存在诸多内容安全隐患,如视频内容篡改、插播、恶意传播等,严重阻碍了网络视频业务可持续性发展。视频指纹基于视频内容且与内容唯一对应,为解决视频内容安全监控提供了一种新思路和方法,但实时性问题是目前视频指纹应用于视频内容监控的一个技术瓶颈。本课题以"提高视频指纹获取的实时性和检索的快速性"为核心目标,以视频指纹技术为研究对象,利用其在内容识别方面的独特优势,从视频指纹生成、指纹存储、指纹检索等角度深入研究视频指纹高效提取和检索的核心算法及关键技术,并通过实验仿真平台分析和评估相应指标的实际性能。拟在基于最小顶点覆盖原理的关键帧快速抉择方法、基于视觉感知的指纹高效提取算法、指纹关联存储结构以及指纹快速检索机制等方面有所突破和创新。项目研究成果将为网络视频业务内容安全实时监控的大规模部署和推广提供理论依据和技术支撑。
中文关键词: 视频内容安全;视频指纹;指纹检索;关键帧;指纹存储
英文摘要: Videos have become the main media information in present network, but there are still many security risks during the transmission of videos, i.e. illegal video; tampered video etc. These security risks will hinder the development of video service. Video fingerprinting technology is content-based and has strong robust and discrimination, so it gives a novel method for the video content security monitoring. Although video fingerprinting has great advantage in content identification, real-time problem will be a technological bottleneck for applying to video content security monitoring. The core object of this project is to real-time obtain and rapid retrieval the video fingerprint. Thus this project will study some core algorithms and key technologies about video fingerprinting technology, i.e. video fingerprinting extraction, storage, retrieval. We will analyze and assess the performance of these algorithms and technologies by a simulation platform. This project will provide following breakthroughs and innovations: choice mechanism of video key frame based on minimum vertex cover principle; extraction algorithm of video fingerprinting based on visual perception theory; storage structure of video fingerprinting; fast retrieval algorithm in fingerprinting database. Finally, the above achievements will provide theore
英文关键词: video content security;video fingerprinting;fingerprinting retrieval;key frames;fingerprinting storage