摘要
针对银行监控系统中监控视频信息冗余度高、浏览效率低等问题,提出一种基于多层关联目标跟踪的视频浓缩算法。该算法首先通过基于聚类的目标检测算法获取运动目标,其次将检测结果通过多层关联目标跟踪,获取运动目标的运动轨迹,并将目标图片和视频信息结构化保存在本地,最后将这些目标重新组合整理,回贴到背景图片上得到浓缩视频。实验结果表明,通过该算法得到的浓缩视频,能够在不丢失视频信息的前提下,减少存储空间,节省硬件成本,缩短浏览时间,提高相关人员的工作效率。
Many video in the bank surveillance system have the high redundancy and low efficiency browse problem, we propose a method of video concentrate based on multi-hierarchical data association object tracking. Firstly, to get moving target by object detection based on clustering. Secondly,to obtain object trajectories from detections using the method of multi-hierarchical data association object tracking,and then save the object frame and its video information into local file. Finally,the moving objects and video background are reassembled into the concentrated video. The experimenlal results show that the method proposed in this paper can reduce the storage space and hardware cost without losing important video information. It is also can save more time for some related person and improve their work efficiency.
出处
《软件导刊》
2016年第10期169-172,共4页
Software Guide
关键词
视频浓缩
目标跟踪
目标检测
背景建模
最小代价图
智能监控
Video Condensation
Object Tracking
Object Detection
Background Modeling
Minimum Cost Graph G
Intelligent Monitoring