Low complexity algorithm for the extraction of vehicular traffic variables

We present an alternative technique for acquisition of traffic variables, using less computer resources and less processing time, by using a computer vision algorithm that process only non redundant information. The algorithm allows estimating vehicular mean speed and vehicular volume in 2 or 3 lanes roads. It distinguishes vehicles from the background using a single line of the image and estimates the mean speed of the vehicles by using two lines of the image. This minimizes both, the quantity of information that has to be processed, and the total processing time. The algorithm also uses a spatio-temporal transformation, allowing more time for the detection process, therefore reducing hardware requirements.