Performance Evaluation of Video Based Obstacle Detection Algorithm for Railway Level Crossing

초록

This paper presents a performance-tested obstacle detecting algorithm using image processing which is adequate for the level-crossing environment. The proposed algorithm accurately detects objects by analyzing their trajectories in order to minimize the rate of false alarms due to shadows and changes in lighting in the environment of the level crossing. It also outperforms the existing image-sensing process algorithm by complementing the drawbacks appearing the method of background image modeling with small amount of calculation through post-processing, such as object tracking. The performance of the algorithm with the obstacle image sensing device was tested on video during crossing operations.