火力与指挥控制
火力與指揮控製
화력여지휘공제
Fire Control & Command Control
2015年
9期
138-142
,共5页
智能视频监控%异常行为检测%分层背景%Mean-shift算法
智能視頻鑑控%異常行為檢測%分層揹景%Mean-shift算法
지능시빈감공%이상행위검측%분층배경%Mean-shift산법
intelligent video surveillance%detection of abnormal behavior%layering background%Mean-shift algorithm
针对复杂环境下视频监控滞后严重难以满足预警的问题,提出一种基于计算机视觉技术的复杂环境下异常行为检测算法。该算法首先利用分层背景模型对环境中的行人目标进行检测;其次采用改进的Mean-shift算法对运动目标进行实时跟踪,并在跟踪结果的基础上进行运动目标的异常行为判断。实验结果表明:该算法可以建立复杂环境准确的背景模型,有效地对入侵目标进行检测;可以对跟踪目标进行实时跟踪,有效地对逆行和逗留进行检测。
針對複雜環境下視頻鑑控滯後嚴重難以滿足預警的問題,提齣一種基于計算機視覺技術的複雜環境下異常行為檢測算法。該算法首先利用分層揹景模型對環境中的行人目標進行檢測;其次採用改進的Mean-shift算法對運動目標進行實時跟蹤,併在跟蹤結果的基礎上進行運動目標的異常行為判斷。實驗結果錶明:該算法可以建立複雜環境準確的揹景模型,有效地對入侵目標進行檢測;可以對跟蹤目標進行實時跟蹤,有效地對逆行和逗留進行檢測。
침대복잡배경하시빈감공체후엄중난이만족예경적문제,제출일충기우계산궤시각기술적복잡배경하이상행위검측산법。해산법수선이용분층배경모형대배경중적행인목표진행검측;기차채용개진적Mean-shift산법대운동목표진행실시근종,병재근종결과적기출상진행운동목표적이상행위판단。실험결과표명:해산법가이건립복잡배경준학적배경모형,유효지대입침목표진행검측;가이대근종목표진행실시근종,유효지대역행화두류진행검측。
Considering the problem that traditional video monitoring tasks under complex environments is too lag to meet the early warning, a detection algorithm based compute vision for abnormal behavior is proposed. Firstly,pedestrian targets are detected under the layering background model. Secondly, the pedestrian targets are tracked by the improved Mean-shift algorithm. And the algorithm will detect whether the moving target’s behavior is abnormal or not. The experiments result that this algorithm can build real-time background model of complex environments to detect invading targets effectively and can track moving targets timely to detect the abnormal behavior of retrograding and sojourn.