Study On Airport Runway Incursion Prevention Based On Computer Vision

The runway serves as the start and end facilities for aircraft operations,so its safety cannot be underestimated.With the increasing traffic flow at airports,runway incursions have occurred frequently,which has a huge impact on the safety of civil aviation.Research on preventing runway incursion is of urgency.Therefore,a computer vision-based method to prevent airport runway incursion is proposed.The main research work includes:Firstly,taking into account that airport surveillance video images are easily affected by the weather and environment,the computer vision-based image enhancement method under low visibility in fog and low brightness at night is studied.Through an efficient dark channel image haze removal method and an adaptive image brightness enhancement method,the airport surveillance videos under low visibility and low brightness scenarios are improved without increasing the time complexity of proposed algorithm.Secondly,an airport surface objects tracking method based on the Deep Sort framework is constructed.The detector framework based on YOLO with an average precision of 34.3on MS COCO dataset can detect airport surface objects efficiently.The tracker predicts the trajectory through Kalman filtering,introduces Mahalanobis distance and cosine distance as features,and matches the airport surface target detected by the detector with the trajectory through cascade matching and Hungarian matching.The simulation results show that the method can effectively track the target on the airport surface,MOTA reaches 0.90,and the method efficiency is up to 23 fps,which can meet the requirements of real-time monitoring.Then,considering that the aircrafts tracking alone cannot characterize the overall situation of airport.Therefore,the key nodes of the airport surveillance are determined,and the airport surface model is constructed.On the basis of targets tracking method,the possible pavement states and changes of airport surface targets are abstracted through runway situation modeling based on finite automata machine.The establishment of the state transition table can judge the current situation and provide information to the airport surface targets effectively combining with the runway status lights.Finally,the runway incursion prevention system has been constructed,developed and tested.The system displayed the content of the airport surface surveillance camera with simple configuration and brief user interface.The system can set the runway status lights automatically without manual operation,so it will not increase the load of the air traffic controllers and the staff of airport.It can monitor the runway incursions in real time and increase the runway safety level.