摘要
针对ViBe(visual background extractor)算法中经常出现的“鬼影”现象,提出一种融合帧间差分法和ViBe算法的运动目标检测算法。首先采用帧间差分法和ViBe算法对视频序列连续三帧分别做运动目标检测;然后对两个相互独立的检测结果进行逻辑运算并更新背景模型;最后对运算结果依次执行开运算、闭运算,从而得到准确的运动目标区域。选择3种常见环境下的视频序列进行试验研究,结果表明我们提出的这种算法可以有效去除“鬼影”,具有良好的鲁棒性和实时性。
In response to the ghosting phenomenon frequenting in ViBe(visual background extractor)algorithm,a moving object detection algorithm was proposed by integrating frame difference with ViBe.Firstly,the frame difference method and ViBe algorithm were used to detect the moving target of video sequence for three consecutive frames respectively,and then the two independent detection results were logically calculated,with the background model updated.Finally,the operation results were successively performed on the open operation and the closed operation,so as to obtain the accurate moving target area.Video sequences in three common environments were selected for the experimental research.The results show that this algorithm boasts good robustness and real-time performance,capable of effectively removing the ghosting.
作者
陈宁
范英豪
白冰
CHEN Ning;FAN Yinghao;BAI Bing(School of Mechanical and Energy Engineering,Zhejiang University of Science and Technology,Hangzhou 310023,Zhejiang,China;Zhejiang Key Laboratory of Food Logistics Equipment Technology Research,Hangzhou 310023,Zhejiang,China)
出处
《浙江科技学院学报》
CAS
2020年第1期32-37,共6页
Journal of Zhejiang University of Science and Technology
基金
国家自然科学基金项目(61273240)
浙江省科技计划项目(2019C04025)
关键词
ViBe算法
帧间差分法
背景模型
逻辑运算
ViBe algorithm
frame difference method
background model
logical operations
