摘要
设计了一种新的基于自动阈值分割的混合模型运动检测算法。将单高斯背景模型中求解门限阈值的思想引入,实现了混合模型门限的自适应调整和运动目标的分割。为提高运动检测的准确性,使用相邻三帧差法生成的时域运动前景掩模对检测结果进行修正。实验结果表明,混合模型算法不仅能够清晰准确地获得运动目标的完整信息,而且较好地消除了噪声的干扰。
This paper designs a novel hybrid moving object detecting algorithm based on auto threshold segmentation. Through introducing the idea of uni-Gaussian model thresholding method, the proposed hybrid algorithm succeeds in realizing the auto threshold adjustment and foreground object segmentation. To ensure the accuracy of moving object detection, a temporal foreground mask is generated by three-frame differencing, and then applied to correcting the detected result. Experimental results show that this algorithm can obtain clear and complete information of the moving object, and eliminate noise fundamentally for the scene at the same time.
出处
《计算机工程与应用》
CSCD
2012年第31期171-174,190,共5页
Computer Engineering and Applications
关键词
运动目标检测
自动阈值分割
时域前景掩模
混合模型算法
moving object detection auto threshold segmentation temporal foreground mask hybrid algorithm
