A typical shadow-eliminating algorithm is presented. Algorithms for vehicle detection, edge detection and shadow detection are also presented. If the shadow is not separated from the vehicle, the next traffic control processing, like vehicle classification, tracking and speed calculation, will be difficult. Our approach for shadow processing runs in real time in low-cost hardware. The knowledge of the heading of roads and date and time can be efficient used in the algorithm. First we detect moving vehicles in a sequence of image, hi this way. we need not process the whole image, which saves computation time. Then we separate shades from vehicles using knowledge including results of edges detection, road heading, etc. The results of the algorithm are also given in the final of the paper, which shows that this algorithm is quite promising.