Linkage detection method and system for highway appearance and internal diseases

Abstract

The invention relates to a linkage detection method and system for highway appearance and internal diseases, and the method comprises the steps: synchronously collecting a pavement appearance image and a highway internal electromagnetic response signal in real time, and recording a time-space synchronization label; performing real-time apparent disease detection on the pavement apparent image by adopting a target recognition algorithm, outputting disease types, further extracting disease feature information, and judging the severity of the disease; based on a mapping relation between disease types and disease severity and an internal disease depth range established by priori knowledge, determining the detection depth of underground diseases as the detection target depth of a radar so as to dynamically adjust the antenna frequency combination of the radar, obtain an electromagnetic response signal of a highway internal structure, and convert the electromagnetic response signal into a radar image; extracting electromagnetic reflection characteristics; and performing feature matching and data fusion by adopting a feature matching algorithm to generate a comprehensive disease image and an analysis report. Compared with the prior art, the method has the advantages of improving the precision and efficiency of road disease detection and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of highway disease detection, and in particular to a method and system for linkage detection of highway surface and internal diseases. Background Art

[0002] Highway defect detection is an important part of highway maintenance. Its main purpose is to timely discover and evaluate the defects of highway pavement and internal structure through detection technology, providing a scientific basis for highway maintenance and repair. Traditional detection methods are generally divided into two parts: apparent defect detection and internal defect detection. Apparent defect detection mainly uses image acquisition equipment to obtain surface images of the road surface and uses image processing technology to identify defects such as cracks and potholes; internal defect detection uses equipment such as ground penetrating radar to obtain underground structure information and analyze the distribution of internal defects. However, the existing technology has the following problems:

[0003] 1) Separation of surface and internal defect detection: Traditional methods typically perform surface and internal defect detection separately, lacking a linkage mechanism. This separate approach results in inefficient comprehensive road defect detection and makes it difficult to establish a correlation between surface and internal defects.

[0004] 2) Inadequate detection depth and accuracy: The frequency of ground-penetrating radar (GPR) significantly impacts detection depth and accuracy. High-frequency radar waves have higher resolution but shallower penetration depth, while low-frequency radar waves have greater penetration depth but lower resolution. In existing detection equipment, GPR frequency selection is typically fixed, making it impossible to dynamically adjust detection depth based on surface damage characteristics. This results in insufficient detection accuracy for deeper damage.

[0005] 3) Data Fusion Difficulties: Surface imagery and internal radar data differ in time and space. Surface imagery typically offers high resolution and rich texture information, but its coverage is limited. Internal radar data can provide detailed information about underground structures, but its resolution is lower and is significantly affected by environmental factors. Existing technologies make it difficult to effectively fuse and align these data, hindering comprehensive disease analysis.

[0006] Therefore, a linked detection method is urgently needed to solve the above problems. Summary of the Invention

[0007] The purpose of the present invention is to provide a method and system for detecting highway surface and internal defects, which can realize the linkage detection of highway surface defects and internal hidden defects.

[0008] The purpose of the present invention can be achieved by the following technical solutions:

[0009] A method for detecting surface and internal defects of a highway comprises the following steps:

[0010] Real-time synchronous acquisition of road surface images and electromagnetic response signals of the highway's internal structure, and recording of spatiotemporal synchronization tags. The electromagnetic response signals are obtained by underground radar scanning when the damage occurs.

[0011] Using a target recognition algorithm to perform real-time surface disease detection on the road surface image, output the type of disease, and further extract disease feature information to determine the severity of the disease;

[0012] Based on the mapping relationship between the type and severity of the damage and the depth range of the internal damage established by prior knowledge, the detection depth of the underground damage is determined as the radar detection target depth. The radar antenna frequency combination is dynamically adjusted to re-acquire the electromagnetic response signal of the internal structure of the highway under different antenna frequency combinations, convert it into a radar image, and extract the electromagnetic reflection characteristics;

[0013] Based on the electromagnetic reflection characteristics and disease feature information, combined with the spatiotemporal synchronization tags, a feature matching algorithm is used to perform feature matching and data fusion to generate a comprehensive disease image and analysis report.

[0014] Furthermore, the target recognition algorithm is trained using the YOLOv10 model.

[0015] Furthermore, before the target recognition algorithm performs real-time apparent disease detection, it also includes preprocessing the pavement surface image, and the preprocessing steps include: denoising, enhancement and normalization operations, wherein the denoising operation is performed using a Gaussian filter, and the enhancement operation is performed using automatic exposure compensation and image enhancement algorithm, and the pixel values ​​of the pavement surface image are normalized to the range of 0-1.

[0016] Furthermore, the mapping relationship is:

[0017] When the disease type is network cracks:

[0018] When the severity of the disease is mild, the corresponding detection depth of underground disease is 0.2m.

[0019] When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 0.5m.

[0020] When the severity of the disease is severe, the corresponding detection depth of underground diseases is 1m;

[0021] When the damage type is reflective cracks:

[0022] When the severity of the disease is mild, the corresponding detection depth of underground diseases is 1m.

[0023] When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 2m.

[0024] When the severity of the disease is severe, the corresponding detection depth of underground diseases is 3m;

[0025] When the disease type is pits and grooves:

[0026] When the severity of the disease is mild, the corresponding detection depth of underground disease is 0.5m.

[0027] When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 2-3m.

[0028] When the severity of the disease is severe, the corresponding detection depth of underground diseases is 5m;

[0029] For loose diseases:

[0030] When the severity of the disease is mild, the corresponding detection depth of underground diseases is 2m.

[0031] When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 5m.

[0032] When the severity of the disease is severe, the corresponding detection depth of underground diseases is 8m;

[0033] When the damage type is subsidence:

[0034] When the severity of the disease is mild, the corresponding detection depth of underground diseases is 2m.

[0035] When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 5m.

[0036] When the severity of the disease is severe, the corresponding detection depth of underground diseases is 10m.

[0037] Furthermore, when the type of damage is a network of cracks, the step of extracting the damage characteristic information and determining the severity of the damage includes:

[0038] For the network crack disease identified by the target recognition algorithm, the skeleton of the network crack is extracted using the improved Zhang-Suen thinning algorithm;

[0039] Based on the skeleton of the network of cracks, performing skeleton connected domain analysis to estimate the total length of the network of cracks;

[0040] Based on the skeleton of the mesh cracks, the width of the mesh cracks is calculated by interpolation of the normal direction and the contour boundary;

[0041] Based on the skeleton of the network of cracks, the main direction algorithm is used to estimate the direction and area of ​​the network of cracks;

[0042] Determining the severity of the network of cracks based on the total length and width of the network of cracks;

[0043] The total length, width and area of ​​the network cracks are used as characteristic information of the network cracks.

[0044] Furthermore, the radar's target detection depth and frequency satisfy the following relationship:

[0045]

[0046] Where δ is the depth of the detected target, c is the speed of light, f is the frequency, and ε r is the relative dielectric constant of the medium.

[0047] Furthermore, when the detection depth of the underground disease is a deep disease, the low-frequency antenna group is switched to perform directional focus scanning to obtain the electromagnetic response signal of the internal structure of the highway.

[0048] When the detection depth of the underground disease is a shallow disease, the high-frequency antenna group is switched to perform high-density profile scanning to obtain the electromagnetic response signal of the internal structure of the highway.

[0049] Furthermore, before extracting the electromagnetic reflection features, the radar image is preprocessed, including: denoising, gain control and signal enhancement, wherein a digital filtering method is used for denoising and an environmental noise compensation method is used for signal enhancement.

[0050] Furthermore, the comprehensive disease image includes a two-dimensional image and a three-dimensional model, wherein the two-dimensional image displays the type and location of the disease, and the three-dimensional model displays the depth and three-dimensional shape of the disease.

[0051] The present invention also provides a detection system according to the above-mentioned method for detecting highway surface and internal defects, comprising:

[0052] Image acquisition module: used to collect road surface images;

[0053] Ground Penetrating Radar Module: This module includes multiple transmitting and receiving antennas at different frequencies. It is used to acquire electromagnetic response signals from the internal structure of the highway based on a dynamically adjusted radar antenna frequency combination, convert these signals into radar images, and extract electromagnetic reflection features.

[0054] Data processing module: includes a disease identification unit, a radar frequency adjustment unit, and a data fusion unit. The disease identification unit is used to detect apparent diseases in real time, further extract disease feature information, determine the severity of the disease, and determine the detection depth of underground diseases based on prior knowledge;

[0055] The radar frequency adjustment unit is used to dynamically adjust the antenna frequency combination of the radar according to the detection depth of the underground disease;

[0056] The data fusion unit is used to perform feature matching and data fusion on the electromagnetic reflection features and disease feature information using spatiotemporal synchronization tags and feature matching algorithms to generate a comprehensive disease image and analysis report;

[0057] Display module: used to display the comprehensive disease image and analysis report.

[0058] Compared with the prior art, the present invention has the following beneficial effects:

[0059] (1) The present invention uses the types and severity of apparent diseases detected by the optical camera as prior knowledge to dynamically adjust the frequency domain combination of the ground penetrating radar. This not only solves the problems of insufficient accuracy and low efficiency caused by the separation of apparent and underground disease detection and the lack of a linkage mechanism, but also can achieve accurate detection of hidden underground diseases at different depths, effectively improving the accuracy and efficiency of highway disease detection, and providing comprehensive and accurate disease information for road management and maintenance.

[0060] (2) This invention introduces surface damage information as prior knowledge for the dynamic adjustment of GPR parameters for the first time, breaking through the limitations of traditional fixed-band scanning. It dynamically adjusts the frequency combination of multi-frequency GPRs, achieving precise detection of underground damage at different depths while more accurately identifying and locating damage bodies of different depths and sizes, while improving detection accuracy and efficiency. When detecting shallow targets, higher-frequency radar waves are used to achieve higher resolution; when detecting deep targets, lower-frequency radar waves are switched to increase the detection depth, effectively improving the accuracy and versatility of detecting different types of damage.

[0061] (3) The present invention uses an optical camera and a ground-penetrating radar to collaboratively collect data, and dynamically selects the scanning frequency band combination of the ground-penetrating radar according to the apparent disease characteristics to achieve focused detection of underground diseases within the target depth range; uses spatiotemporal synchronization tags and feature matching algorithms to align the optical image and the ground-penetrating radar image data B-scan in a unified spatiotemporal coordinate system, and fuses them to generate a comprehensive disease image, thus solving the key problem of low spatiotemporal alignment accuracy of multi-source data. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] Figure 1 Schematic diagram of the method flow of the present invention;

[0063] Figure 2 Schematic diagram of the corresponding relationship between the apparent disease characteristics and the depth range of hidden diseases and the optimal radar frequency combination of the present invention;

[0064] Figure 3This is a schematic diagram of a comprehensive disease image of the present invention;

[0065] Figure 4 This is a structural diagram of the system integrated on the inspection vehicle of the present invention;

[0066] Among them, ① is the image acquisition module, ② is the ground penetrating radar module, ③ is the data processing module, and ④ is the display module. DETAILED DESCRIPTION

[0067] The present invention is described in detail below with reference to the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0068] This embodiment provides a method for detecting surface and internal defects of highways, aiming to solve the problem of insufficient accuracy and low efficiency caused by the separation of surface and underground defect detection and the lack of a linkage mechanism in the existing technology. The linkage detection method is integrated into a linkage detection system, which is installed on a detection vehicle, such as Figure 4 As shown, it includes image acquisition module 1, ground penetrating radar module 2, data processing module 3, and display module 4. The data processing module 3 includes a disease identification unit, a radar frequency adjustment unit, and a data fusion unit. The data processing module 3 is embedded in the data processing server embedded in the inspection vehicle, equipped with NVIDIA Jetson AGX Orin module and built-in parallel computing architecture. Specifically, combined with Figure 1 and Figure 4 As shown, the execution steps of the linkage detection method and system include the following:

[0069] Step S1: Image acquisition and apparent disease identification

[0070] In this embodiment, the inspection vehicle travels normally on the highway at a speed of 60 km / h. Through the synchronous controller, a high-resolution camera (resolution ≥ 20 million pixels) and a ground-penetrating radar are simultaneously activated to obtain the surface image of the road surface and the electromagnetic response signal inside the highway, respectively, and record the spatiotemporal synchronization tags, where the electromagnetic response signal is obtained by scanning at the default lowest frequency when there is no disease.

[0071] Image acquisition module 1 comprises a four-camera array mounted on the front and rear stabilized gimbals of the inspection vehicle. Using a binocular stereo vision algorithm, it reconstructs the road surface model. It covers a 3.5-meter inspection width and supports blur-free image capture at speeds between 0 and 80 km / h. The cameras are arranged at multiple angles (0°, 45°, 90°, and 135°) to ensure comprehensive coverage of the road surface, preventing missed defects due to viewing angle limitations. The cameras also feature automatic exposure compensation and image enhancement to adapt to varying lighting and weather conditions. High-resolution cameras capture road surface images in real time. Each camera acquires images according to preset parameters (such as a 35mm focal length, f / 4 aperture, 1920×1080 resolution, and 1 / 1000 exposure time) to ensure image clarity and quality.

[0072] The collected road surface images first undergo preprocessing, including denoising, enhancement, and normalization. Automatic exposure compensation and image enhancement algorithms are used to reduce the impact of lighting conditions and reflected light on image quality. A Gaussian filter (with a kernel size of 5×5) is used to remove random noise from the image. Image pixel values ​​are normalized to a range of 0–1 for ease of subsequent processing.

[0073] The preprocessed pavement surface image is transmitted to the Defect Identification Unit in Data Processing Module 3. This unit deploys a TensorRT-accelerated YOLOv10 object recognition algorithm model to perform real-time image detection, identifying common defects such as cracks (network cracks and reflective cracks), potholes, looseness, and subsidence. The output is formatted as JSON and contains structured data such as defect type, bounding box coordinates, affected area percentage, and confidence level, which is stored on the server for subsequent analysis. The unit also determines the severity of the defect, providing a basis for subsequent dynamic adjustments.

[0074] In this embodiment, the disease recognition unit identifies the common highway surface disease of network cracks based on the preprocessed road surface image, and uses the improved Zhang-Suen thinning algorithm to extract the skeleton of the network cracks to ensure that the width of the network cracks is a single pixel. Based on the connected domain analysis of the skeleton, the total length of the network cracks is estimated. The width of the cracks is calculated by interpolation of the normal direction and the contour boundary, and the direction of the cracks is estimated using the main direction algorithm. At the same time, the area of ​​the network cracks is calculated, and the above feature information is combined to classify the disease. According to the output feature information of the network cracks (length 50cm, width 2mm), its severity is marked as "mild". The recognition results are stored in the form of structured data in JSON format.

[0075] Step S2: Prediction of underground disease depth range

[0076] Based on the depth distribution patterns of defects in the "Highway Subgrade Design Code" (JTG D30-2015), the severity levels are classified as mild, moderate, and severe. The possible depth range of internal defects is predicted based on the defect type and severity. For example, network cracks, caused by deterioration of the asphalt surface material, can range in depth from 0.2m to 1m; subsidence defects, caused by uneven subgrade settlement, can reach depths of 2m to 10m.

[0077] The system uses prior knowledge to establish a mapping relationship between the type of disease, the severity of the disease, and the depth range, and quickly locks the detection target depth of the ground penetrating radar. The mapping relationship table is as follows:

[0078] Web-like cracks: mild (0.2m), moderate (0.5m), severe (1m)

[0079] Reflective cracks: mild (1m), moderate (2m), severe (3m)

[0080] Potholes: Mild (0.5m), Moderate (2-3m), Severe (5m)

[0081] Loose: Mild (2m), Moderate (5m), Severe (8m)

[0082] Subsidence: Mild (2m), Moderate (5m), Severe (10m)

[0083] According to the predicted depth range, the radar frequency adjustment unit provides parameter support for dynamically adjusting the frequency combination of the ground penetrating radar by preparing the corresponding antenna frequency combination.

[0084] In this embodiment, since the identified network of cracks is "mild", according to the mapping relationship, the corresponding internal disease depth range is predicted to be 0.2m.

[0085] Step S3: Dynamic adjustment and detection of ground penetrating radar frequency

[0086] In this embodiment, the ground-penetrating radar module 2 is 15-20 cm above the ground surface and is primarily used to detect defects in the structural layers and base layers of highways. The radar's frequency range is 200 MHz to 2000 MHz, with an effective detection range of 0-15 meters. It includes transmitting and receiving antennas capable of emitting radar waves at different frequencies, specifically four lower-frequency antennas (200 MHz, 400 MHz, 600 MHz, and 800 MHz) and four higher-frequency antennas (1000 MHz, 1200 MHz, 1500 MHz, and 2000 MHz). It supports MIMO (Multiple-Input Multiple-Output) operating mode.

[0087] According to the radar wave attenuation model formula:

[0088]

[0089] Where δ is the detection depth, c is the speed of light, f is the frequency, ε r is the relative dielectric constant of the medium. Accordingly, the radar frequency adjustment unit dynamically selects the appropriate antenna frequency combination according to the depth range of the disease.

[0090] When no apparent defects are detected in step S1, the radar maintains a low frequency of 200MHz to collect underground data. When shallow defects (such as reticular cracks) are detected, the high-frequency antenna group is enabled for high-density profile scanning, and high-frequency antennas (1500MHz, 2000MHz) are preferred to obtain higher resolution; when deep defects (such as severe looseness or subsidence) are detected, the low-frequency antenna group is switched to, for example, a low-frequency antenna (400MHz, 600MHz) is selected to ensure sufficient detection depth. The correspondence between the apparent defect characteristics and the depth range of hidden defects and the optimal radar frequency combination is shown in the figure. Figure 2 shown.

[0091] Based on the adjusted frequency combination, GPR Module 2 uses an FPGA (Field Programmable Gate Array) to achieve nanosecond pulse triggering, ensuring data integrity at high speeds and capturing electromagnetic response signals from the road's internal structure. Using high-precision signal processing technology, GPR Module 2 converts the electromagnetic response signals into radar images.

[0092] The acquired radar images undergo preprocessing, including denoising, gain control, and signal enhancement, before extracting the electromagnetic reflection characteristics of the damage. Digital filtering and ambient noise compensation techniques are used for denoising and signal enhancement, respectively, to ensure the clarity and resolution of the radar images.

[0093] In this embodiment, the ground-penetrating radar module 2 switches to a 2000 MHz high-frequency mode in the crack area and transmits electromagnetic waves underground through the transmitting antenna to obtain higher-resolution images of the damage. The receiving antenna captures the electromagnetic response signals from the road's internal structure, converts them into radar images, and then preprocesses the radar images. The radar images show low electromagnetic reflection intensity in the crack area and good waveform continuity. The electromagnetic reflection characteristics of the cracks extracted from the radar images are: reflection intensity -20 dB and a continuous waveform.

[0094] Step S4: Data fusion and comprehensive analysis

[0095] The data fusion unit is based on the acquisition results of the camera and ground penetrating radar and the spatiotemporal synchronization tags (GPS timestamp + IMU posture data). It uses the SIFT (Scale Invariant Feature Transform) algorithm to extract the geometric feature points in the apparent image and simultaneously extracts the strong reflection area features (electromagnetic reflection features) in the radar image B-scan.

[0096] The RANSAC (random sample consensus) algorithm is used to perform feature matching of multi-source data with a matching error of ≤0.1m, and finally an RGB optical image is superimposed on a pseudo-color map of radar reflection intensity.

[0097] Generate comprehensive disease images and analysis reports based on the fused data, and analyze the location, type, depth and severity of the disease.

[0098] The display module 4 displays a comprehensive disease image and analysis report. The comprehensive disease image includes: a two-dimensional image showing the location and type of the disease, and a three-dimensional model showing the depth and three-dimensional shape of the disease; the comprehensive analysis report indicates the disease status and treatment suggestions.

[0099] In this embodiment, the data fusion unit uses spatiotemporal synchronization tags and SIFT feature matching algorithm to align the apparent image data and radar data on the spatiotemporal axis. The apparent image position of the crack matches the image position in the radar image B-scan, confirming that the crack depth is 0.2m. Figure 3 As shown, the corresponding two-dimensional image and three-dimensional model are displayed on the display module 4, and the comprehensive analysis report indicates that "moderate network cracks are accompanied by slight foundation cracks, and milling and re-paving are required."

[0100] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0101] It will be understood by those skilled in the art that the embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code. The solutions in the embodiments of the present invention may be implemented in various computer languages, for example, the object-oriented programming language Java and the interpreted scripting language JavaScript.

[0102] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowcharts and / or block diagrams, as well as combinations of processes and / or blocks in the flowcharts and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowcharts and / or block diagrams. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0103] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0104] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0105] Although the preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present invention.

[0106] Obviously, those skilled in the art may make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if such changes and modifications fall within the scope of the claims and their equivalents, the present invention is intended to include such changes and modifications.

Claims

1. A method for detecting surface and internal defects of highways, characterized in that: The following steps are involved: Real-time synchronous acquisition of road surface images and electromagnetic response signals of the highway's internal structure, and recording of spatiotemporal synchronization tags. The electromagnetic response signals are obtained by underground radar scanning when the damage occurs. Using a target recognition algorithm to perform real-time surface disease detection on the road surface image, output the type of disease, and further extract disease feature information to determine the severity of the disease; Based on the mapping relationship between the type and severity of the damage and the depth range of the internal damage established by prior knowledge, the detection depth of the underground damage is determined as the radar detection target depth. The radar antenna frequency combination is dynamically adjusted to re-acquire the electromagnetic response signal of the internal structure of the highway under different antenna frequency combinations, convert it into a radar image, and extract the electromagnetic reflection characteristics; Based on the electromagnetic reflection characteristics and disease feature information, combined with the spatiotemporal synchronization tags, a feature matching algorithm is used to perform feature matching and data fusion to generate a comprehensive disease image and analysis report.

2. A method for detecting surface and internal defects of a highway according to claim 1, characterized in that: The target recognition algorithm is trained using the YOLOv10 model.

3. The method for detecting surface and internal defects of a highway according to claim 1, characterized in that: Before the target recognition algorithm performs real-time apparent disease detection, it also includes preprocessing the pavement surface image. The preprocessing steps include: denoising, enhancement and normalization operations, wherein the denoising operation is performed using a Gaussian filter, and the enhancement operation is performed using automatic exposure compensation and image enhancement algorithm, and the pixel values ​​of the pavement surface image are normalized to the range of 0-1.

4. The method for detecting surface and internal defects of a highway according to claim 1, characterized in that: The mapping relationship is: When the disease type is network cracks: When the severity of the disease is mild, the corresponding detection depth of underground disease is 0.2m. When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 0.5m. When the severity of the disease is severe, the corresponding detection depth of underground diseases is 1m; When the damage type is reflective cracks: When the severity of the disease is mild, the corresponding detection depth of underground diseases is 1m. When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 2m. When the severity of the disease is severe, the corresponding detection depth of underground diseases is 3m; When the disease type is pits and grooves: When the severity of the disease is mild, the corresponding detection depth of underground disease is 0.5m. When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 2-3m. When the severity of the disease is severe, the corresponding detection depth of underground diseases is 5m; For loose diseases: When the severity of the disease is mild, the corresponding detection depth of underground diseases is 2m. When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 5m. When the severity of the disease is severe, the corresponding detection depth of underground diseases is 8m; When the damage type is subsidence: When the severity of the disease is mild, the corresponding detection depth of underground diseases is 2m. When the severity of the disease is moderate, the corresponding detection depth of underground diseases is 5m. When the severity of the disease is severe, the corresponding detection depth of underground diseases is 10m.

5. A method for detecting highway surface and internal defects in a linked manner according to claim 4, characterized in that: When the type of disease is a network of cracks, the steps of extracting disease characteristic information and determining the severity of the disease include: For the network crack disease identified by the target recognition algorithm, the skeleton of the network crack is extracted using the improved Zhang-Suen thinning algorithm; Based on the skeleton of the network of cracks, performing skeleton connected domain analysis to estimate the total length of the network of cracks; Based on the skeleton of the mesh cracks, the width of the mesh cracks is calculated by interpolation of the normal direction and the contour boundary; Based on the skeleton of the network of cracks, the main direction algorithm is used to estimate the direction and area of ​​the network of cracks; Determining the severity of the network of cracks based on the total length and width of the network of cracks; The total length, width and area of ​​the network cracks are used as characteristic information of the network cracks.

6. The method for detecting highway surface and internal defects according to claim 1, characterized in that: The radar's target detection depth and frequency satisfy the following requirements: Where δ is the depth of the detected target, c is the speed of light, f is the frequency, and ε r is the relative dielectric constant of the medium.

7. The method for detecting highway surface and internal defects according to claim 1, characterized in that: When the detection depth of the underground disease is deep disease, switch to the low-frequency antenna group to perform directional focus scanning to obtain the electromagnetic response signal of the internal structure of the highway. When the detection depth of the underground disease is a shallow disease, the high-frequency antenna group is switched to perform high-density profile scanning to obtain the electromagnetic response signal of the internal structure of the highway.

8. The method for detecting surface and internal defects of a highway according to claim 1, characterized in that: Before extracting the electromagnetic reflection features, the radar image is preprocessed, including: denoising, gain control and signal enhancement, wherein a digital filtering method is used for denoising and an environmental noise compensation method is used for signal enhancement.

9. The method for detecting highway surface and internal defects according to claim 1, characterized in that: The comprehensive disease image includes a two-dimensional image and a three-dimensional model, wherein the two-dimensional image displays the type and location of the disease, and the three-dimensional model displays the depth and three-dimensional shape of the disease.

10. A detection system for the method for detecting highway surface and internal defects in a linked manner according to any one of claims 1 to 9, characterized in that: include: Image acquisition module (1): used to acquire road surface images; Ground penetrating radar module (2): comprising a plurality of transmitting antennas and receiving antennas of different frequencies, for acquiring electromagnetic response signals of the internal structure of the highway according to a dynamically adjusted radar antenna frequency combination, converting the signals into radar images, and extracting electromagnetic reflection features; Data processing module (3): including a disease identification unit, a radar frequency adjustment unit and a data fusion unit, wherein the disease identification unit is used to detect apparent diseases in real time, further extract disease feature information, determine the severity of the disease, and determine the detection depth of underground diseases based on prior knowledge; The radar frequency adjustment unit is used to dynamically adjust the antenna frequency combination of the radar according to the detection depth of the underground disease; The data fusion unit is used to perform feature matching and data fusion on the electromagnetic reflection features and disease feature information using spatiotemporal synchronization tags and feature matching algorithms to generate a comprehensive disease image and analysis report; Display module (4): used for displaying the comprehensive disease image and analysis report.

Images