摘要
ViBe算法是一种快速高效的背景建模算法,但该算法在运动目标检测过程中会产生鬼影。针对ViBe算法中鬼影消除缓慢的问题,结合多个场景的交通视频提出一种通过连续两帧前景背景像素时域变化来判断鬼影像素点并消除的方法,该方法加快了鬼影的消除速度。同时,对于视频拍摄场景中的背景噪声,采用了对前景图进行开闭操作去除小像素点以及对目标区域的空洞进行填充处理。实验表明,改进的ViBe算法能够加快鬼影的消除,并且与帧差法以及混合高斯建模算法相比,前景检测效果更精确。
ViBe algorithm is an effective and quick background modeling algorithm. During the moving target detection,the ViBe will produce the ghost and eliminate slowly. In order to solve the problems caused by the execution of ViBe algorithm,this paper proposed an improved algorithm for the traffic video. It judged the ghost pixel by the change of two consecutive frames in time domain,which could speed up the ghost to eliminate. In addition,to solve the noise in the complex background,it applied a morphology post-processing with opening operation and closing operation to remove the noise,then filling the holes in the target zone to highlight the prospect target. Experiments show that the improved algorithm can absorb the ghost quickly and detect the foreground better compared with frame-difference and GMM.
出处
《计算机应用研究》
CSCD
北大核心
2017年第5期1565-1567,1571,共4页
Application Research of Computers
基金
国家自然科学基金资助项目(51107093)
关键词
运动目标检测
ViBe算法
帧差法
鬼影消除
moving object detection
ViBe algorithm
frame-difference algorithm
ghost elimination
