Research On Spectral Characteristics Of Typical Foreign Materials On Airport Runways And Their Detection Applications

With the rapid development of civil aviation,the traffic on airport runways is increasing day by day.Statistics show that the probability of safety accidents during takeoff,landing,and taxiing stages is high,and foreign objects on airport runways are the main hidden danger affecting flight safety.Foreign objects on airport runways may pierce aircraft tires or be sucked into engines,causing damage to the aircraft and even threatening the safety of people’s lives and property.Traditional runway foreign object detection mostly relies on human visual inspection,such as using patrol cars or walking to inspect with nets,which is inefficient and inaccurate.Currently,vision-based runway foreign object detection methods are receiving attention and becoming a research hotspot.However,morphology-based foreign object detection methods are affected by factors such as video quality,resulting in a high false alarm rate.In this paper,based on the material properties of foreign objects,hyperspectral imaging detection technology is used to realize foreign object detection and identification with the joint information of the image spectrum.While detecting the target,the material type of the foreign object can also be identified,facilitating risk assessment and disposal.The specific work is as follows:(1)Statistical analysis of foreign objects on airport runways.Detailed statistical information on runway foreign objects is lacking in existing public data.The research team relied on the Xinzheng Airport Flight Area Runway Management Department to compile a record of runway foreign objects over the past three years,collect physical samples of runway foreign objects,and analyze their size,quantity,and material characteristics to summarize the typical foreign object material types that are high in occurrence and high in risk.(2)Determination and analysis of typical foreign bodies on airport runways and construction of hyperspectral database.The composition of typical foreign bodies on the runway was determined by modern material testing and analysis methods,and the spectral absorption characteristic mechanism was demonstrated based on chemical bonds and functional groups.The hyperspectral data of typical foreign objects were collected in visible to near-infrared bands(350 nm-1000 nm).Based on image dimension and spectral dimension data annotation,the typical foreign body image and reflectivity curve database of airport runway were constructed.(3)Determination and analysis of typical foreign object composition on airport runways and construction of a hyperspectral dataset.Modern material testing and analysis methods were used to determine the composition of typical foreign objects on runways,based on the spectral absorption characteristics mechanism of chemical bonds and functional groups.Hyperspectral data was collected for typical foreign object targets in the visible to near-infrared band(350 nm-1000 nm).Data annotation was carried out based on the image dimension and spectral dimension to construct an airport runway typical foreign object image and reflectance curve dataset.(4)Image segmentation and material spectral recognition of typical foreign objects on airport runways.In the image segmentation and extraction stage,a runway foreign object extraction method combining U-net segmentation and morphological operations was proposed,using runway foreign object image dimension labels to achieve foreign object target segmentation and extraction in the runway background,forming a region of interest(ROI),and improving the F1 score by 3.6% to describe the extraction accuracy.In the foreign object material recognition stage,based on the spectral dimension dataset,a BP neural network-based ROI material discrimination detection was implemented,with an average detection accuracy of 77.2% and a false alarm rate of only 0.95%.