Ground penetrating radar roadbed disease target detection method based on convolutional neural network

Abstract

The invention discloses a ground penetrating radar roadbed disease target detection method based on a convolutional neural network. The method comprises the steps of obtaining original image data of ground penetrating radar simulation and actual collection; performing corresponding preprocessing on the original image data to obtain ground penetrating radar preprocessed images; marking and storingtarget information in the original and preprocessed images; constructing a PASCAL VOC data set by the ground penetrating radar image and the marking information file, and dividing the PASCAL VOC dataset into a training set, a verification set and a test set; dynamically setting an anchor frame parameter initial value in combination with the marked target frame aspect ratio; and then adopting a Cascade R-CNN network built by training set training and verification set fine tuning to obtain a convolutional network model, evaluating the performance of the network model by using the test set, andfinally realizing accurate and rapid detection of the roadbed disease target of the ground penetrating radar. The method does not depend on artificial identification, has strong generalization ability, and can realize rapid and accurate detection of roadbed disease targets.

Description

technical field [0001] The invention relates to the field of ground penetrating radar signal processing, in particular to a ground penetrating radar roadbed disease target detection method based on a convolutional neural network. Background technique [0002] Subgrades are vital to road and railway. Due to construction conditions, geographical environment, climate, vehicle driving and other reasons, there are many diseases on the road. The surface and shallow defects of highways and railways are easy to observe and detect, but the diseases at the subgrade position are not easy to be found. If they are not dealt with in time and effectively, they will affect the use of highways and railways and seriously threaten the safety of drivers. As a non-destructive, high-accuracy, fast-efficiency, and adaptable detection technology, ground-penetrating radar replaces the original lossy and non-destructive detection methods, and is widely used in roadbed disease detection projects. ...

AI Technical Summary

Problems solved by technology

Although the above method can detect the target, it relies on human analysis and identification, and requires technicians with rich experience and prior knowledge to understand and master a large number of structural characteristics of roadbed disease targets, which contains many subjective factors; it takes a lot of energy and time. The detection efficiency is low; and due to manual operation, the obtained feature parameters and feature representations are less, and the generalization ability is lacking, resulting in low detection accuracy and affecting the judgment of roadbed diseases

[0005] With the development of machine learning in recent years, combined with the representation of different targets in GPR B-Scan image data, for example, circular targets in different media present a hyperbolic structure with polar differences, and square targets present a double-sided structure on both sides. Curves, structures with parallel lines in the middle, etc., according to their shape, polarity and other performance characteristics, machine learning methods are used to realize automatic detection of roadbed disease targets. Although manual extraction of target recognition and detection is not required, due to algorithm design and other reasons, other Accurate detection is still not possible for complex roadbed environments

Images