Research On Target Detection And Tracking,in The System Of Aircraft Ground Traction Collision Aviodance

At present, aircraft ground collision avoidance relies mainly on scene monitoring system, and realizes monitoring and alarm by installing equipments on the airport scene, but not on the tow truck. Thus, the contents of this paper will have good research and application prospects. Since the camera is moving because it's installed on the tow truck, therefore, we need to study the moving target detection and tracking in the condition of dynamic scene. The main work of this article is as follows:1, This paper presents a moving target detection method based on SURF, the algorithm's computational complexity is low, real-time performance is good, and it can accurately detect the feature points on the moving target, and carry on stable tracking.Also, the algorithm have a certain stability when the light changes or there is noise.2, The different feature points on the same object have approximative velocity, we use cluster analysis based on data connectivity to cluster the key points to get the different key points in the background and moving objects, then all of the key points in the background are filtered out.3, For using only Camshift algorithms during the target tracking is susceptible to interference by the large area of similar background problems, this paper proposes a combined Camshift and Kalman filter target tracking algorithms. Using Kalman filter instead of Camshift when there is large area of similar background interference, to a certain extent, solved the interference problem.4, According to the proposed motion detection and tracking algorithm, designed and implemented a prototype collision avoidance system for aircraft ground movement, and a preliminary test in the practical environment. From the experimental results, we can see that in the general environmental conditions, applying this design to complete the aircraft ground collision avoidance systems can achieve better results.