对象和关键帧相结合的监控视频摘要提取方法  

Extraction Method of Surveillance Video Synopsis Combines Objects and Keyframes

作　　者：王亚沛[1] 李仁旺[1] 刘翔[1] 

机构地区：[1]浙江理工大学机械与自动控制学院,浙江杭州310018

出　　处：《工业控制计算机》2015年第3期11-13,共3页Industrial Control Computer

基　　金：国家自然科学基金(51475434)

摘　　要：参考了基于对象和基于关键帧的视频摘要技术,阐述了一种基于对象的视频摘要技术的主要方法流程,并在传统基于对象方法的基础上融入了改进的关键帧提取方法,以单个运动对象为目标进行视频空间上的压缩,通过实验验证,在不遗漏关键信息的前提下大幅度地减少了视频冗余信息,使用户可以在短时间内更加快速便捷地阅读视频监控内容,找出有用信息。同时提出了一种更为合理的视频摘要合成流程,为视频摘要领域的深入研究提供一定的参考。This paper refers to the video synopsis technology based on objects and keyframes,it describes the main method of a kind of video synopsis process based on objects,and into the improvement method of keyframes extraction under the traditional mothed based on objects,compress the space of video with single motion object,by experimental verification,it reduces video redundant information as possible under the premise of not missing critical information,allows the user to more quickly and convenient read the video surveillance content in a short time,and find out useful information.The article also presents a more reasonable synthesis process of video synopsis,provides certain reference of in-depth study in the field of video synopsis.

关 键 词：视频摘要 对象 关键帧 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]