Vision-based vehicle detection and tracking method for forward collision warning in automobiles

We describe a vision-based vehicle detection and tracking method for forward collision warning in automobiles. The approach is based on a set of edge-based constraint filters that assist in the segmentation of vehicles from background clutter. The detected vehicles are then tracked using a combination of distance based matching, sum-of-square-of-difference in intensity (SSD) and edge density of detected vehicle regions. The computational load for tracking is minimized using a vehicle-clustering algorithm. Experimental results are presented to illustrate the performance of the algorithm.