Vehicle detection using running Gaussian average and Laplacian of Gaussian in the nighttime

Abstract

In this paper, we propose a nighttime vehicle detection algorithm using RGA (Gunning Gaussian Average) and LoG (Laplacian of Gaussian). At the first stage, background could be estimated using RGA from CCTV input image. At the next stage, 2-D LoG function applied to input image. And then "AND" operator could be applied between each other. Finally, threshold of headlight brightness could be supported to the output of AND operation. As the results of simulations using recoded real video signal, it is shown that proposed algorithm is useful for detection vehicles in the nighttime. © 2012 Springer-Verlag.