Improving Method For Identification Accuracy Of Tunnel Lining Radar Nondestructive Detection Based On Image Features

With the rapid development of my country’s economy,my country’s basic industry construction is also constantly expanding and progressing,especially in the construction of highways,a large amount of capital and manpower are invested.The highway tunnel is a construction that is more difficult to construct in the construction of highways.Therefore,the quality inspection of highway tunnels that are already under construction or already completed has put forward more qualified specifications and more accurate targets.In the tunnel quality inspection,the tunnel lining is a highly concealed inspection item.Due to the complex internal structure,the traditional borehole inspection method has a greater damage to the lining structure,and the accuracy of discrimination is low,which cannot provide accurate detection.Reference content,so geological radar detection has become a commonly used detection method in highway tunnels.Ground penetrating radar uses the principle of electromagnetic wave reflection to scan and discriminate the internal structure of concrete.It only needs to analyze from the images in the collected data to identify the internal reinforcement and voiding of the lining.This method is simple to operate,does not require destruction tests,and can greatly guarantee the integrity of the construction,so it is used as an effective non-destructive testing method.In this paper,in the long-term practice of using geological radar,combined with physical experiments,image processing and other methods to study the identification accuracy of the tunnel lining void image,to improve the accuracy of the current inspection work.the follows are some research results:(1)Build a physical model based on the principle of ground penetrating radar to obtain the airborne radar detection imageBy referring to data and theoretical study,we can understand the basic theory of electromagnetic wave emission and the principle of ground penetrating radar detection,and analyze and summarize the influencing factors of ground penetrating radar.On this basis,first use the construction materials on site to simulate the construction of a cuboid coagulation block.In the coagulation block model,combined with the construction parameters,four vacant areas with different shapes and regular shapes are constructed inside.Separated by a certain interval and in the same horizontal center position.Then use the CO1760 integrated ground penetrating radar instrument produced by Impluse Radar to detect the physical model and obtain the inspection image.(2)Identify the void area in the detection image based on the principle of gray distributionBased on the acquired radar detection image,it is found that the essence is a grayscale image.By comparing the physical model detection image with the initial support lining detection image,the image characteristics of the radar detection image void area are summarized.A series of processes such as image enhancement,image edge detection,adaptive threshold segmentation,and removal of small connected domains were successfully extracted from the physical model radar detection image using Matlab software to successfully extract the voided area image.(3)Propose four kinds of measurement indexes of radar detection image recognition accuracyCombined with the research of previous scholars,the resolution of the radar is divided into two indexes,vertical accuracy and horizontal accuracy,and the detection accuracy of the radar is obtained based on the plane error rate and the buried depth error rate.In this paper,the void area of the detected image is obtained based on the principle of gray distribution,and then the horizontal projection and vertical projection features of the gray level are used to obtain the parameters of the length of Ferret path and the coordinates of the center point of the void area.The obtained voiding information is compared with the position parameters in the physical model,and four indexes,namely,discrimination rate,plane error rate,buried depth error rate and Ferret diameter error rate,are proposed to measure the recognition accuracy of radar detection images.(4)Application of engineering examplesThrough a series of model tests and verifications,the identification accuracy of the GPR image is finally obtained.With the help of the engineering conditions of the Longnaoke tunnel of the Baihetan reconstruction project,the vacancy situation in the tunnel lining was successfully detected,and the location and size of the disease were counted through the analysis of image feature indexes.