Airfield pavement disease foreign matter detection method based on fusion of multiple convolutional neural networks

Abstract

The invention provides an airport pavement disease and foreign matter detection method based on fusion of multiple convolutional neural networks. The method comprises the steps of collecting airport pavement disease and foreign matter images; constructing an airport pavement disease foreign matter database for training a neural network; building a target detection algorithm YOLOv3 and a Mask R-CNN (Convolutional Neural Network); the hyper-parameters of the convolutional neural network are adjusted until convergence and an error loss value meet requirements, the network weight parameters at the moment are stored, and training of the YOLOv3 and Mask R-CNN convolutional neural networks is completed; the trained YOLOv3 and Mask R-CNN convolutional neural networks are fused, and an intelligent segmentation model of the airport pavement disease and foreign matter pixel level is constructed; inputting a test image into the stored model, and outputting a segmentation result of the diseases and the foreign matters of the airport pavement; and carrying out statistics on pixels of the image corresponding to the mask of the segmentation result, and outputting semantic information of airport pavement diseases and foreign matters. The method has better robustness and generalization ability, and can improve the segmentation precision and efficiency of airport pavement disease foreign matters.

Description

technical field [0001] The invention belongs to the technical field of non-destructive detection of foreign matter on airport pavement, and in particular relates to a method for detecting foreign matter on airport pavement based on optimization and fusion of multiple convolutional neural networks. Background technique [0002] Airport pavement diseases mainly include cracks, potholes, peeling and other forms, among which cracks are one of the main forms of diseases, and are the early manifestations of most structural diseases, which will cause great safety hazards during aircraft take-off and landing. Foreign objects on airport pavement (Foreign Object Debris, FOD) refer to any object that appears in an inappropriate position in the airport, such as fragments falling off the pavement, missing parts of aircraft, etc. Foreign matter on the road surface directly affects aviation safety. Foreign matter on the road surface may be sucked into the engine, causing damage to the engi...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the above problems, the present invention provides a method for detecting foreign objects on airport pavement based on the fusion of multiple convolutional neural networks, which can fully consider the small size of foreign objects on airport pavement and the complex environment of the image background, and integrate various convolutional neural networks. Based on the advantages of the neural network algorithm, the recognition accuracy and noise resistance of the algorithm are improved, so as to ensure the common and simultaneous detection of foreign objects on the airport pavement, and to make up for the defects of manual inspection

Images