Research And Application On Airport Runway Defect Detection Algorithm

Airport runways carry the take-off and landing of the aircraft day and night,and there will be defects such as foreign matters and cracks in the use process.These defects will bring potential safety hazards to the aircraft take-off and landing.It is of great significance to detect the defects of the airport runway regularly and timely and maintain the runway for the civil aviation flight safety.Airport runway defect detection technology is the key to detect runway defects.The research on this technology has both theoretical and practical value.At present,it is difficult to collect a large number of defect image samples in a short period of time on the airport runway in operation,and the existing detection algorithm has low accuracy under the interference of complex pavement texture.At the same time,in the practical application,the inspection of vehicle terminal is adopted,which has real-time requirements for the defect detection with high risk level.Therefore,this paper studies the defect detection algorithm according to the above situation,and the specific research contents are as follows:First of all,in order to expand the defect image sample data,this paper studies the construction of airport runway image data set,optimizes and improves the image generation algorithm based on the generated countermeasure network,and combines the method of random oversampling to generate defect image samples,so as to build the airport runway data set.Secondly,aiming at the real-time requirement of defect detection with high risk level in practical application,this paper proposes a method of airport runway defect detection based on Tiny Yolo,which uses a smaller model to obtain a higher detection speed in case of loss of certain detection accuracy.Next,the algorithm of airport runway defect detection under complex pavement texture is studied,and the algorithm of airport runway defect detection based on waveilbp texture feature is proposed.Firstly,the algorithm solves the problem that the quality of runway image is degraded due to the uneven illumination,and enhances the runway image.The waveilbp texture feature is proposed by combining the spectrum method and the statistical method.The wavelet decomposition is used to highlight the frequency-domain feature of the defect edge,and the mean method is used to optimize the LBP texture feature.Then,based on the texture features,clustering algorithm is used to detect the defects.Compared with gray co-occurrence matrix,waveilbp texture features have higher detection accuracy.Finally,applying the above research results,this paper designs and realizes the airport runway defect detection system,and completes the design and implementation of data transmission,defect detection,monitoring alarm and system security module.The test results show that the system achieves the design purpose.In conclusion,the algorithm studied in this paper has a high defect detection accuracy and can meet the needs of airport runway defect detection application.