MV-GPRNet: Multi-View Subsurface Defect Detection Network for Airport Runway Inspection Based on GPR

Abstract

The detection and restoration of subsurface defects are essential for ensuring the structural reliability of airport runways. Subsurface inspections can be performed with the aid of a robot equipped with a Ground Penetrating Radar (GPR). However, interpreting GPR data is extremely difficult, as GPR data usually contains severe clutter interference. In addition, many different types of subsurface defects present similar features in B-scan images, making them difficult to distinguish. Consequently, this makes later maintenance work harder as different subsurface defects require different restoration measures. Thus, to automate the inspection process and improve defect identification accuracy, a novel deep learning algorithm, MV-GPRNet, is proposed. Instead of traditionally using GPR B-scan images only, MV-GPRNet utilizes multi-view GPR data to robustly detect regions with defects despite significant interference. It originally fuses the 3D feature map in C-scan data and the 2D feature map in Top-scan data for defect classification and localization. With our runway inspection robot, a large number of real runway data sets from three international airports have been used to extensively test our method. Experimental results indicate that the proposed MV-GPRNet outperforms state-of-the-art (SOTA) approaches. In particular, MV-GPRNet achieves F1 measurements for voids, cracks, subsidences, and pipes at 91%, 69%, 90%, and 100%, respectively.

Defect segmentation with local embedding in industrial 3D point clouds based on transformer

“…Tong et al [31] proposed a joint learning, which effectively combines multi-view features and classifiers, reducing redundant information and improving the accuracy of point cloud classification. Li et al [10] used fused multi-view 2D and 3D feature maps as input to train CNN for defect classification and localization of airport runways. Zhou et al [12] proposed a method for non-contact defect detection of circuit boards using template matching was developed.…”

“…Due to the significant structural differences between 3D point clouds and 2D images, the disorderly and irregular nature of point clouds can lead to few 3D defect detection methods using them as input. To address this problem, many researchers have converted 3D point clouds into regular 2D images by multi-angle projection [10][11][12][13] to facilitate processing using 2D image detection methods. However, using multi-angle projection can become challenging when dealing with complex objects and industrial scenes with occlusion problems.…”

A Comprehensive Review of Conventional and Deep Learning Approaches for Ground-Penetrating Radar Detection of Raw Data

“…The paper is tested using a real-world runway dataset from three international airports. The test results show that the proposed method is superior to the state-of-the-art (SOTA) method and achieves a high detection rate in a variety of defect detections [96]. The network structure of [96] is shown in Figure 16.…”

“…The test results show that the proposed method is superior to the state-of-the-art (SOTA) method and achieves a high detection rate in a variety of defect detections [96]. The network structure of [96] is shown in Figure 16. In the same year, M. Zhong et al proposed a hybrid 3D electromagnetic full-wave inversion method for the reconstruction of underground detection targets.…”