A Traffic Camera Calibration Method Based on Multi-rectangle

Abstract : As distance detection is frequently needed in modern traffic management, camera calibration has a significant influence on the accuracy of distance detection. The traditional calibration methods usually need a calibration target, and the exact distance of calibration points has to be measured. In this paper, according to the camera imaging model, we put forward a calibration method based on multi-rectangle, which constructs several rectangles with the mark lines of the traffic road. Without measuring the internal or external parameters of the camera or the exact distance of calibration point, we only need the side lengths of rectangles and the image coordinates of the rectangular vertexes to establish the video image distance conversion model. For improving the accuracy of conversion, we use multiple rectangles to determine the coordinates of the vanishing points and modify the coordinates of the rectangular vertexes. Experimental results show that our method is more accurate when compared with the mainstream methods.