融合目标速度变化机制的视频摘要生成模型  

A Video Synopsis Generation Model Incorporating Object Speed Change Mechanism

作　　者：牛嘉丰 石蕴玉[1] 刘翔[1] 李任斯 NIU Jiafeng;SHI Yunyu;LIU Xiang;LI Rensi(College of Electronic and Electrical Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)

机构地区：[1]上海工程技术大学电子电气工程学院,上海201620

出　　处：《电子科技》2022年第9期30-36,共7页Electronic Science and Technology

基　　金：国家自然科学基金(81101105);上海市自然科学基金(19ZR1421500);文化部科技创新项目(2015KJCXXM19)。

摘　　要：视频摘要是监控视频浏览和存储的有效技术。在压缩比的限制下,现有的视频摘要生成方法容易导致重排目标碰撞和时序错乱。针对该问题,文中提出一种融合目标速度变化机制的视频摘要生成模型。在目标重排的能量函数中,除目标起始位置变量之外还加入了目标速度变量,从而能够同时改变目标的起始位置和速度,避免碰撞和时序错乱问题。随后使用马尔科夫链蒙特卡罗随机采样算法求解能量函数的最优值,得到了目标重排方案的最优解。实验表明,在压缩率相同的情况下,相较于其他方法,该模型生成的摘要视频中的目标碰撞和时序错乱的问题较少。Video synopsis is an effective technique for surveillance video browsing and storage.Existing video synopsis generation methods are prone to cause rearrangement objects collision and chronological disorder under the limitation of compression ratio.To solve this problem,a video synopsis generation model incorporating object speed change mechanism is proposed in this study.In the energy function of object rearrangement,object speed variable is added in addition to object start position variable,so that object start position and speed can be changed simultaneously to avoid collision and chronological disorder problems.Then the Markov chain Monte Carlo random sampling algorithm is used to solve the optimal value of the energy function,and the optimal solution of the target rearrangement scheme is obtained.Experiments show that when the compression rate is the same,compared with other methods,the summary video generated by the proposed model has fewer problems with target collision and timing disorder.

关 键 词：视频摘要 目标速度 蒙特卡罗采样 目标碰撞 目标重排 监控视频 能量函数 泊松编辑 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]