The Detection Of Roof Foreign Object Based On Laser Scanning System

Railway as China’s leading transport infrastructure and the main means of transportation during Spring Festival, the passenger and cargo transport system is in a stage of rapid development, but also accompanied by a huge security operation pressure. The presence of roof foreign object is a security risk to train operation, such as leading damage to inherent equipment on trains, causing the train can not work fine and so on. But the detection and localization of foreign bodies on the roof of the locomotive is still a difficulty problem, there is not a complete system in China and abroad can automatically complete the foreign body detection of locomotive roof.To improve the quality of locomotive running, this paper proposes a foreign object detection method based on laser scanning combined with point cloud processing technologies. In order to obtain the scanning data, a three-dimensional laser scanner LMS400 was used to scan the roof. The scanner is not susceptible to light and other external interference. Then the point cloud could be reconstructed by scanner calibration and speed correction. Data obtained by the sensor are in polar coordinates,including distance and angle information. In order to obtain real three-dimensional point cloud, we add velocity factor in the forward direction of the train, so the point cloud becomes the real one in the process of actual movement. After using improved ICP(Iterative Closest Point) algorithm for point cloud registration, improved background subtraction was adopted to extract the foreign objects.In this paper, the pre-simulation was conducted with MATLAB, after the realization of all algorithms and meeting the testing requirements, the algorithms will be converted into C code. The results show that, the robust ICP registration algorithm based on correntropy has both high registration accuracy and fast convergence. Also, foreign object can be accurately detected. Finally, the paper summarizes the process and made a number of areas for improvement, to make a follow-up study task outlook.