3-dimension traffic reconstruction by automatic camera calibration under homogenous fog
Hongjun Song, Yuanyuan Gao · 2014
A novel algorithm for 3-Dimension traffic reconstruction by automatic and dynamic camera calibration based on dark channel prior in homogenous fog weather condition is presented in this article. Camera fixed in the middle of road is calibrated by scene distance information implicated in homogenous fog weather condition. Unlike other works in 3-D traffic scene reconstruction, this proposed algorithm is executed by camera calibration and the traffic model that only includes road plane and vehicles in motion. Priori information (prior distance information marked by manual in traffic scene and vehicle length) is not available here. Painted lines are not neglected in the proposed model because sometime there are no traffic lanes in un-structured road. Four main steps compose our algorithm. At the end of this paper, calibration results and 3-D reconstruction results will be given. Comparison with other algorithm in MATLAB verifies the effectiveness of our proposed algorithm.