摘要
为了实现旋转摄像头下运动目标的监控,提出了一种动态背景下提取运动目标的方法.在全局运动估计过程中,首先提取当前帧的特征点并将其离散化,然后以特征点为中心在当前帧和前一帧之间进行块匹配,其后用3σ准则除去误差块,从而得到全局运动估计量,并且在VC平台上实现了该算法.
In order to detect moving object by a rotating camera in video surveillance, a moving object extraction method based on dynamic background is proposed. First, the feature points are extracted and discretized. Then feature points are used as the center for block - matching between consecutive frames and frame difference is used after compensation. And then, the rule is used to remove errors. Experiments prove that the method is efficient;.
出处
《佳木斯大学学报（自然科学版）》
CAS
2009年第2期174-176,共3页
Journal of Jiamusi University：Natural Science Edition
基金
航空基金(NO20060752006)
关键词
动态背景
特征点提取
背景匹配
dynamic background
feature points extraction
background matching
