Real-time background estimation of traffic imagery using group-based histogram
Kai‐Tai Song, Jen Chao Tai · 2008
This paper presents a novel background estimation method for a vision-based traffic monitoring system using a single Gaussian scheme. An algorithm of group-based histo-gram (GBH) is proposed to build the background Gaussian model of each pixel from traffic image sequences. This algorithm features improved robustness against transient stop of foreground objects and sensing noise. Furthermore, the method features low computational load, thus meets the real-time requirements in many practical applications. The proposed method has been applied to a vision-based traffic parameter estimation system to segment moving vehicles from image sequences. Given degraded compressive traffic images from on-line internet cameras, the image processing system successfully detect various vehicles in the traffic imagery. Practical experimental results demonstrate that traffic flow can be measured in real time with satisfactory accuracy.