摘要
运动车辆进行高效实时检测存在目标轮廓缺失、目标内部有空洞、背景无法实时更新等困难。基于此,提出了改进的三帧间差分法和帧间差分法、背景差分法融合算法,在应用改进三帧间差分法解决目标轮廓不完整、利用基于混合高斯模型建模的背景差分法实现目标背景的实时更新基础上,将两者融合,进行运动车辆检测。实验证明,所提出的算法可以较好地解决运动目标轮廓缺失以及背景无法实时更新的问题。
The efficient and real-time detection of moving vehicles has difficulties such as missing target contours and backgrounds that cannot be updated in real time.Based on this,this paper put forward an improved three-frame difference method and a fusion algorithm based on inter-frame difference method and background difference method.The proposed methods can eliminate incomplete target contour,realize the real-time update of the target background.Experiments prove that the algorithm proposed in this paper can better solve the problem of missing contours of moving targets and the background cannot be updated in real time.
作者
韩帅
Han Shuai(College of Mechanical and Automotive Engineering,Shanghai University of Engineering Science,Shanghai 201620,China)
出处
《农业装备与车辆工程》
2021年第6期88-92,共5页
Agricultural Equipment & Vehicle Engineering
关键词
背景差分法
三帧间差分法
混合高斯模型
运动车辆检测
background difference method
three-frame difference method
mixed Gaussian model
moving vehicle detection
