一种基于层次分析法的视频摘要评价模型  

Evaluation model of video summarization based on analytic hierarchy process

作　　者：栾悉道[1] 谢毓湘[2] 李琛[2] 牛晓[2] Luan Xidao Xie Yuxiang Li Chen Niu Xiao(Dept. of Mathematics ＆ Computer Science, Changsha University, Changsha 410022, China College of Information Systems ＆ Management, National University of Defense Technology, Changsha 410073 , China)

机构地区：[1]长沙学院数学与计算机科学系,长沙410022 [2]国防科学技术大学信息系统与管理学院,长沙410073

出　　处：《计算机应用研究》2016年第12期3832-3835,共4页Application Research of Computers

基　　金：国家自然科学基金资助项目(61571453);湖南省自然科学基金资助项目(14JJ3010);湖南省教育厅重点资助项目(15A020)

摘　　要：是对视频内容进行浓缩的一项技术,对于快速了解视频内容至关重要。如何对视频摘要的效果进行评价是值得研究的一个问题。基于层次分析法构建了视频摘要评价模型,将视频摘要质量作为最终评价目标,以内容合理性和结构合理性作为准则,以内容完整性、特殊重要性、整体流畅性等作为测度层,从而建立了视频摘要评价指标体系。最后,通过对随机生成、基于特写人脸的摘要生成以及融合视音频特征的摘要生成三种算法对所提评价模型进行了实验验证,表明该模型能够有效反映出视频摘要的质量。The goal of video summarization is to summary the video content, which is important to help to grasp the main idea of video content. How to evaluate the summarization effect is a problem to be solved urgently. This paper constructed a video summarization evaluation model based on analytic hierarchy process （AHP） method. In this model, the target was the quality of video summarization, the rule was the content and structure rationality, and the measurement layer was the integrality, the im- portance and the fluency of video content. These constructed the evaluation factor system of video summarization. Finally, by experiments on three different algorithms such as random generation ,feature face-based generation and audio-visual feature fusion generation, the proposed model is proved to be effective in reflecting the quality of video content.

关 键 词：视频摘要 评价模型 层次分析法 

分 类 号：TP391.1[自动化与计算机技术—计算机应用技术]