An Improved Calibration Algorithm for ITS Based On Vision

Abstract

In video surveillance for intelligent traffic management system (ITS), the nonlinear distortion of image will lead to the inaccurate parameter of traffic. This paper presents a new algorithm for global calibrate using information of road marking lines, which is a common feature in traffic surveillance scenes and is easy to obtain. The new algorithm is simple, available, effective, accurate, without knowing camera's interior and exterior parameters, complex setting, and vast storage of data. It also overcomes the limitations of the method based on grid mapping. Experiments proved that the algorithm can be applied not only to traffic surveillance system accurately and effectively, but also to other different industrial solutions based on computer vision.