摘要
Vibe算法是通用的运动检测算法,用视频序列第一帧初始化背景模型,这使得模型的初始化非常迅速,但是当待检测的前景目标在第一帧中出现时,鬼影将伴随着模型的初始化而产生.Vibe算法并未对鬼影做任何特殊处理,以至于Vibe需要花费很长的一段时间才能完成对鬼影的抑制.为了解决这个问题,依赖于Vibe模型本身,提出了一种快速抑制鬼影的算法.当某像素值被Vibe模型判为前景时,将它与邻域像素点模型中的像素值继续比较,当与邻域的匹配达到阈值时,则认定此像素值为鬼影像素值,继而从前景中消去此鬼影点,并重新初始化此像素点的背景模型.实验表明,改进后的算法,与原算法相比,能够迅速的抑制鬼影.
Vibe,a universal background subtraction algorithm,made the initialization very fast by using the first frame to initialize its background model.The only drawback was the moving target in the first frame would introduce an artifact called ghost.As Vibe does nothing extra to eliminate the ghost area,so it took a long process for ghost to fade over time.In order to solve this problem,we proposed a fast eliminating ghost algorithm depending on the Vibe model.When a pixel value was judged as foreground by Vibe,we continued to compare it with its neighboring model.If matching number reached the threshold,we identified the pixel value as ghost,then eliminated this pixel from foreground and re-initialized the pixels background model. Experiments showed that the improved algorithm can quickly absorb the ghost compared with the original algorithm.
出处
《中国计量学院学报》
2013年第4期425-429,共5页
Journal of China Jiliang University
