Research On FOD Detection Algorithm Based On Video Technology For Airport Runway

Since the21st century,airport has been acting as an important platform for air transport withthe rapid development of aviation industry and the expansion of air transport. More and moreattention has been focused on the safety of the airport runway, especially the foreign objects debris(FOD) detection have been significant to the security of civil aviation. The existing FOD detectionsystems abroad adopt millimeter-wave radar and video surveillance technology as the mainachievement for the automatic detection and identification. Due to the huge costs andtechnological monopoly, it is difficult to popularize these systems at domestic airports.This paper provides a low-cost video-based FOD detection project based on binocular visiontechnology for airport runway, equipped on the airport runway patrol car. Firstly, we introduce theproject and the functional modules of video detection systems according to FAA standards.Analysing the shortage of target detection algorithms applied to FOD detection, we prove thefeasibility and advantage using binocular vision measuring technology.To reduce the impact affected by weather conditions during the process of image acquisition,this paper improved Mean Shift filtering based on Retinex for image enhancement. Firstly usingprincipal component analysis (PCA), low contrast image is decomposed into two parts: luminanceand chromatic; secondly applying the improved Mean Shift filtering to get the enhancement ofluminance; at last, restoring the chromatic independently at the same time, and compensating theimage based on the global analysis.In this paper, FOD detection is base on the depth being significantly different from the roadsurface. Firstly, after calibrating the binocular camera, and the homologous images are rectificatedby epipolar rectification, and the two images are mapped by numerical analysis method. Lastly theFOD detection is completed in view field area according to the change in the corresponding pixelmapping, which can overcome the road disturbance in the traditional target detection.Experimental results show that, the FOD detection algorithms mentioned in this paper candetect the FOD effectively even under complicated road conditions, and can effectively enhancethe video image. This project provides new idea for the runway FOD detection system.