Algorithms Study And Software Implementation Of Airport Runway FOD Detection System

FOD (Foreign Object Debris) is a foreign substance which appears on airport runway accidentally and threatens flight safety. The existence of FOD causes great danger and heavy loss to aviation security. At home and abroad, people are paying more and more attention to FOD detection. While several FOD detection systems have been put into use abroad, research on this field is also going deeper in our country. Nowadays the main detection devices lie on radar and camera, and each of them has its own advantages. The FOD detection system discussed in this dissertation uses radar as its primary detection device.The FOD detection system in this dissertation can get data using radar, analyze data with program, and then do image processing and FOD detection, and send information through network. This dissertation focuses on the software part of system, finishing a Visual C++program inside the FOD detection system. With friendly interface, effective data management, and valid processing algorithm, the software is designed to be a mature and stable software product.One main achievement of this dissertation is research on FOD detection algorithm. This dissertation studies traditional camera image processing algorithms, makes improvements according to radar image and finally presents a FOD detection algorithm based on radar image processing. The algorithm includes special denoising, improved background subtraction, thresholding, median filtering, morphology operation, and object segmentation based on flood-fill. With excellent detection accuracy and real-time performance, this algorithm is the core of FOD detection system.Another work of this dissertation is research on FOD recognition algorithm. It talks about popular object recognition algorithms first and brings in recognition algorithm based on unsupervised feature learning. It contains data preparation, feature learning, feature extraction, and image classification. This is a reasonable solution to FOD recognition problem. This dissertation improves the algorithm and finishes the implementation of it, and achieves impressive experimental result. This part of research can be merged into the FOD detection system to improve its function in the future.