Abstract: To satisfy the realtime demand for the traffic parameters in the Intelligent Transportation System, it proposed a new method by combining the framedifference method with the backgrounddifference method for vehicle tracking with the invariance property of the background in a short time. This method was used to complete a rapid background extraction by the method of twice framedifference, and the binary difference image was obtained by the method of backgrounddifference. Morphology operation was introduced to handle the binary image, which can improve the recognition rate to get a relatively complete connected domain. At last, vehicles were tracked, based on the distance between the connected domain of the close frames. Experimental results show that the scheme has good realtime and accuracy performance as well as robustness to meet the demand in the Intelligent Transportation System.

Key words: vehicle tracking; background extraction; framedifference; background extraction based on twice framedifference; background difference