Traffic flow control using artificial vision techniques

In this paper the design and development of an application that aims to detect and estimate the number of vehicles on a road intersection are presented, in order to maximize the traffic light functioning, so that the waiting time depends on the traffic needs. First, a selection process of the interest region is applied to the image sequences, multiplying a mask image with the original image to focus the segmentation in this region. Then, it is segmented by an iterative algorithm, which estimates the background to offset the light intensity variation, it extracts the objects on the road and, through morphological processing, removes the small lines and shapes. Finally, an analysis based on obtained contours areas calculation and addition, which compared with an experimentally obtained rate, determines the road occupation level, and controls the traffic lights status, based on this occupancy level. In experiments with different video sequences the proposed algorithm allows to control the traffic lights status by 95%, adequately.