A method and device for identifying cracks in the base course of asphalt pavement based on ground-penetrating radar

Abstract

This invention discloses a method and device for identifying cracks in the base course of asphalt pavement based on ground-penetrating radar (GPR), belonging to the field of road engineering pavement maintenance technology. The method includes: 1) identifying crack images from radar slices using the YOLOv8n model; 2) extracting A-scan signals from crack regions based on the identification results and classifying time-frequency features using the LightGBM model; 3) clustering the classification results using a "multi-segment clustering + PCA" analysis method to filter out falsely detected cracks. The device is used to execute the above methods. This invention integrates image data and time-frequency features, solving the problem of high false detection rates caused by subjective annotation or objective interference in single identification methods. Experiments show that the false detection rate is reduced from 11.2% to 2.8%, improving the accuracy and engineering applicability of crack identification.

Description

Technical Field

[0001] This invention belongs to the field of road engineering pavement maintenance technology, specifically relating to a method and device for identifying cracks in the base course of asphalt pavement based on ground-penetrating radar. Background Technology

[0002] With the development of road maintenance technology, pavement internal defect identification methods based on ground-penetrating radar (GPR) have gradually become a research hotspot. Existing technologies mainly fall into two categories:

[0003] 1. Deep learning-based image recognition methods:

[0004] Automatic crack detection in radar slice images using convolutional neural networks (CNNs) replaces traditional manual interpretation, solving the problems of low efficiency and high subjectivity in manual processing. However, the performance of this model is affected by the subjectivity of human annotation, as different annotators have different criteria for judging cracks; the limited size of the dataset leads to insufficient generalization ability of the model, resulting in a high false detection rate in practical applications (e.g., the measured false detection rate reached 11.2%).

[0005] 2. Identification method based on the time-frequency characteristics of ground-penetrating radar:

[0006] Time-frequency analysis (such as wavelet transform) can be used to uncover the physical characteristics of cracks (such as electromagnetic wave reflection characteristics), reducing the reliance on image annotation. However, electromagnetic signals are easily affected by objective factors such as material properties, interlayer bonding, and construction quality, leading to excessive model sensitivity; and the models lack stability in complex road environments, making them difficult to apply directly to practical engineering.

[0007] The two types of methods mentioned above are complementary in their recognition mechanisms, but both are limited by a single data dimension: image recognition methods rely on visual features and are easily constrained by the subjectivity of annotation; time-frequency analysis methods rely on physical signals and are easily interfered with by objective environmental factors.

[0008] This results in a high false detection rate and insufficient reliability of existing GPR data automatic identification technology in actual road crack detection, making it difficult to meet engineering needs. Summary of the Invention

[0009] The purpose of this invention is to provide a method and device for identifying cracks in the base course of asphalt pavement based on ground-penetrating radar. This method combines image recognition results with time-frequency feature analysis, which helps to reduce the interference of human subjectivity on recognition accuracy, effectively filter crack image recognition results, and reduce the false detection rate of cracks. This can solve at least one of the technical problems mentioned in the background art.

[0010] To solve the above-mentioned technical problems, the present invention is implemented as follows:

[0011] This invention provides a method for identifying cracks in the base course of asphalt pavement based on ground-penetrating radar, comprising the following steps:

[0012] (1) Crack detection is performed on the radar slice image using the image recognition module to obtain the initial crack region annotation box;

[0013] (2) Based on the location of the marked box in step (1), extract the A-scan signal of the corresponding area in the original ground penetrating radar data;

[0014] (3) Classify the time-frequency characteristics of the A-scan signal using a time-frequency classification model and output the crack prediction results;

[0015] (4) Perform multi-segment clustering and principal component analysis (PCA) on the crack prediction results to filter out falsely detected cracks. The principal component analysis includes:

[0016] Extract the coordinates of the marker points from the crack prediction results;

[0017] The DBSCAN algorithm is used to perform multi-segment clustering of the marked points;

[0018] Principal component analysis is performed on each clustering result to generate crack fitting line segments;

[0019] Vertical crack segments are retained based on angle selection.

[0020] Optionally, in step (1), the image recognition module is implemented using the YOLOv8n model.

[0021] Optionally, step (1) specifically includes:

[0022] Construct a radar slice image dataset and label valid crack samples using a cross-validation mechanism;

[0023] The YOLOv8n model was used to train the dataset and identify cracks.

[0024] Optionally, in step (3), the time-frequency classification model is the LightGBM model, which is used to classify the A-scan time-frequency features of cracks and normal pavement.

[0025] Optionally, in step (4), the multi-segment clustering and principal component analysis further include:

[0026] Convert the radar slice image to the HSV color space and extract the coordinates of the crack prediction markers.

[0027] Principal component analysis is used to model the orientation of clustered point sets, generating fitted line segments distributed along the principal axis;

[0028] Line segments with an angle less than a threshold to the vertical direction are selected as valid cracks.

[0029] The present invention also provides a ground-penetrating radar-based asphalt pavement base course crack identification device for performing the method, comprising:

[0030] Image recognition module: used to detect cracks in radar slices and obtain initial crack region bounding boxes;

[0031] Signal extraction module: used to extract the A-scan signal of the corresponding area in the raw ground penetrating radar data based on the location of the marked box;

[0032] Time-frequency classification module: Classifies the time-frequency features of the A-scan signal using a time-frequency classification model and outputs crack prediction results;

[0033] Clustering and filtering module: Performs multi-segment clustering and principal component analysis on the crack prediction results to filter out falsely detected cracks.

[0034] Optionally, the image recognition module is implemented using the YOLOv8n model.

[0035] Optionally, the time-frequency classification module is implemented using the LightGBM model.

[0036] Compared with the prior art, the advantages of this invention are as follows:

[0037] 1. This invention reduces the false detection rate of cracks from 11.2% for a single image recognition method to 2.8% by integrating ground-penetrating radar image recognition and time-frequency feature analysis, combined with a clustering filtering method of "multi-segment clustering + PCA", significantly improving the reliability of the recognition results.

[0038] 2. This invention uses the YOLOv8n model to perform initial screening of cracks in radar slice images, with an mAP50 of 97.4% and a map50-90 of 64.6%, which can well balance crack identification accuracy and speed; based on the LightGBM model, crack time-frequency feature classification is performed with an accuracy of 92.5%, effectively mining the deep physical features of cracks.

[0039] 3. This invention reduces the subjective influence of the dataset by using a cross-verification annotation mechanism (independent annotation by three technicians + consistent sample screening); it combines the complementary advantages of map (resistant to objective interference) and time-frequency (resistant to subjective interference) to overcome the limitations of a single method in complex road conditions.

[0040] 4. This invention proposes an integrated framework of "image recognition - time-frequency classification - clustering filtering" to achieve full-process automation from raw data analysis and feature extraction to result visualization; it designs a multi-channel radar spatial mapping method (HSV color segmentation + hierarchical dynamic partitioning) to accurately associate crack location with time-frequency signals; and it adopts the DBSCAN+PCA clustering algorithm to adapt to the tortuous distribution characteristics of cracks in the base layer, accurately fit the crack morphology and filter discrete false detection points.

[0041] 5. The method of integrating "map + time and frequency" in this invention filtered out 6 false cracks and 2 cracks that were not effectively filtered, with a false detection rate of only 2.8%, which is 8.4% lower than the previous method. This verifies the potential of the method in engineering applications in road maintenance scenarios. Attached Figure Description

[0042] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort, wherein:

[0043] Figure 1 A flowchart illustrating the method for identifying cracks in the base course of asphalt pavement based on ground-penetrating radar provided in an embodiment of the present invention;

[0044] Figure 2 This is one of the hardware structure diagrams of the electronic device provided in the embodiments of the present invention;

[0045] Figure 3 This is the second schematic diagram of the hardware structure of the electronic device provided in the embodiment of the present invention. Detailed Implementation

[0046] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0047] The terms "first," "second," etc., used in the specification and claims of this invention are used to distinguish similar objects and not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class and the number of objects is not limited; for example, a first object can be one or more. Furthermore, in the specification and claims, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0048] Please see Figure 1 As shown, this embodiment of the invention provides a method for identifying cracks in the base course of asphalt pavement based on ground-penetrating radar, including the following steps:

[0049] Step S1: Detect cracks in the radar slice image using the image recognition module to obtain initial crack region bounding boxes;

[0050] Step S2: Based on the location of the bounding box in Step S1, extract the A-scan signal of the corresponding area from the original ground-penetrating radar data;

[0051] Step S3: Classify the time-frequency features of the A-scan signal using a time-frequency classification model and output the crack prediction results;

[0052] Step S4: Perform multi-segment clustering and principal component analysis on the crack prediction results to filter out falsely detected cracks. The principal component analysis includes:

[0053] Extract the coordinates of the marker points from the crack prediction results;

[0054] The DBSCAN algorithm is used to perform multi-segment clustering of the marked points;

[0055] Principal component analysis is performed on each clustering result to generate crack fitting line segments;

[0056] Vertical crack segments are retained based on angle selection.

[0057] In step S1, the image recognition module is implemented using the YOLOv8n model. The YOLOv8n model has a map50 of 97.4% and a map50-90 of 64.6%, which can well balance the accuracy and speed of crack recognition.

[0058] Step S1 specifically includes:

[0059] Step S11: Construct a radar slice image dataset and label valid crack samples using a cross-validation mechanism;

[0060] Step S12: Use the YOLOv8n model to train the dataset and identify cracks.

[0061] In step S11, the cross-verification mechanism refers to organizing three technicians with experience in radar image interpretation to independently annotate the crack slice image, and selecting cracks that are unanimously identified by the three annotators as valid crack samples. The radar slice image is then exported as a 640... 640 pixels (300 data points) were used to label the radar slices, reducing the subjective influence of the dataset labeling process.

[0062] In step S3, the time-frequency classification model is the LightGBM model, which is used to classify the A-scan time-frequency features of cracks and normal road surfaces.

[0063] This invention uses the LightGBM model as the time-frequency classification model, which can achieve a crack classification accuracy of 92.5%.

[0064] In step S4, multi-segment clustering and principal component analysis further include:

[0065] Convert the radar slice image to the HSV color space and extract the coordinates of the crack prediction markers.

[0066] Principal component analysis is used to model the orientation of clustered point sets, generating fitted line segments distributed along the principal axis;

[0067] Line segments with an angle less than a threshold to the vertical direction are selected as valid cracks.

[0068] Experimental verification

[0069] The method provided by this invention, along with existing radar slicing methods, was used to identify 77 cracks. Experimental results showed that the radar slicing method falsely detected 8 cracks, a false detection rate of 11.2%. In contrast, the method in this application, which integrates "map + time-frequency" analysis, filtered out 6 false cracks, leaving 2 unfiltered, resulting in a false detection rate of only 2.8%, a reduction of 8.4%. In other words, the ground-penetrating radar-based crack identification method using "map + time-frequency" analysis demonstrates good accuracy and engineering adaptability in identifying cracks in the foundation layer, validating its application potential in practical road maintenance.

[0070] The present invention also provides a ground-penetrating radar-based asphalt pavement base layer crack identification device for performing the method, including an image recognition module, a signal extraction module, a time-frequency classification module, and a clustering filtering module.

[0071] The image recognition module is used to detect cracks in the radar slice image and obtain initial crack region bounding boxes. The image recognition module is implemented using the YOLOv8n model.

[0072] The signal extraction module is used to extract the A-scan signal of the corresponding area in the raw ground-penetrating radar data based on the location of the marked box.

[0073] The time-frequency classification module uses a time-frequency classification model to classify the time-frequency features of the A-scan signal and outputs crack prediction results. The time-frequency classification module is implemented using the LightGBM model.

[0074] The clustering and filtering module performs multi-segment clustering and principal component analysis on the crack prediction results to filter out falsely detected cracks.

[0075] See also Figure 2 As shown, this embodiment of the invention also provides an electronic device 700, which includes a processor 710, a memory 709, and a program or instructions stored in the memory 709 and executable on the processor 710. When the program or instructions are executed by the processor 710, they implement the various processes of the above-described embodiment of the method for identifying cracks in the base course of asphalt pavement based on ground penetrating radar and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0076] It should be noted that the electronic devices in the embodiments of the present invention include the mobile electronic devices and non-mobile electronic devices described above.

[0077] Figure 3 A schematic diagram of the hardware structure of an electronic device to implement an embodiment of the present invention.

[0078] The electronic device 700 includes, but is not limited to, components such as: radio frequency unit 701, network module 702, audio output unit 703, input unit 704, sensor 705, display unit 706, user input unit 707, interface unit 708, memory 709, and processor 710.

[0079] Those skilled in the art will understand that the electronic device 700 may also include a power supply (such as a battery) for supplying power to various components. The power supply may be logically connected to the processor 710 through a power management system, thereby enabling functions such as managing charging, discharging, and power consumption through the power management system. Figure 3 The electronic device structure shown does not constitute a limitation on the electronic device. The electronic device may include more or fewer components than shown, or combine certain components, or have different component arrangements, which will not be elaborated here.

[0080] It should be understood that, in this embodiment of the invention, the input unit 704 may include a graphics processing unit (GPU) 7041 and a microphone 7042. The GPU 7041 processes image data of still images or videos obtained by an image capture device (such as a camera) in video capture mode or image capture mode. The display unit 706 may include a display panel 7061, which may be configured in the form of a liquid crystal display, an organic light-emitting diode, etc. The user input unit 707 includes a touch panel 7071 and other input devices 7072. The touch panel 7071 is also called a touch screen. The touch panel 7071 may include a touch detection device and a touch controller. Other input devices 7072 may include, but are not limited to, physical keyboards, function keys (such as volume control buttons, power buttons, etc.), trackballs, mice, and joysticks, which will not be described in detail here. The memory 709 can be used to store software programs and various data, including but not limited to applications and operating systems. The processor 710 may integrate an application processor and a modem processor, wherein the application processor mainly handles the operating system, user interface, and applications, and the modem processor mainly handles wireless communication. It is understandable that the aforementioned modem processor may not be integrated into the processor 710.

[0081] This invention also provides a readable storage medium storing a program or instructions. When the program or instructions are executed by a processor, they implement the various processes of the above-described method for identifying cracks in the base course of asphalt pavement based on ground penetrating radar, and achieve the same technical effect. To avoid repetition, these will not be described again here.

[0082] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes computer-readable storage media, such as computer read-only memory (ROM), random access memory (RAM), magnetic disk, or optical disk.

[0083] This invention also provides a chip, which includes a processor and a communication interface. The communication interface is coupled to the processor. The processor is used to run programs or instructions to implement the various processes of the above-described embodiment of the asphalt pavement base layer crack identification method based on ground penetrating radar, and can achieve the same technical effect. To avoid repetition, it will not be described again here.

[0084] It should be understood that the chip mentioned in the embodiments of the present invention may also be referred to as a system-on-a-chip, system chip, chip system, or system-on-a-chip, etc.

[0085] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0086] Furthermore, it should be noted that the scope of the methods and systems in the embodiments of the present invention is not limited to performing functions in the order shown or discussed, but may also include performing functions substantially simultaneously or in the reverse order, depending on the functions involved. For example, the described methods may be performed in a different order than described, and various steps may be added, omitted, or combined. In addition, features described with reference to certain examples may be combined in other examples.

[0087] The embodiments of the present invention have been described above with reference to the accompanying drawings. However, the present invention is not limited to the specific embodiments described above. The specific embodiments described above are merely illustrative and not restrictive. Those skilled in the art can make many other forms under the guidance of the present invention without departing from the spirit and scope of the claims, and all of these forms are within the protection scope of the present invention.

Claims

1. A method for identifying cracks in the base course of asphalt pavement based on ground-penetrating radar, characterized in that, Includes the following steps: (1) Crack detection is performed on the radar slice image using the image recognition module to obtain the initial crack region annotation box; (2) Based on the location of the marked box in step (1), extract the A-scan signal of the corresponding area in the original ground penetrating radar data; (3) Classify the time-frequency characteristics of the A-scan signal using a time-frequency classification model and output the crack prediction results; (4) Perform multi-segment clustering and principal component analysis on the crack prediction results to filter out falsely detected cracks. The principal component analysis includes: Extract the coordinates of the marker points from the crack prediction results; The DBSCAN algorithm is used to perform multi-segment clustering of the marked points; Principal component analysis is performed on each clustering result to generate crack fitting line segments; Vertical crack segments are retained based on angle filtering. Multi-segment clustering and principal component analysis further include: Convert the radar slice image to the HSV color space and extract the coordinates of the crack prediction markers. Principal component analysis is used to model the orientation of clustered point sets, generating fitted line segments distributed along the principal axis; Line segments with an angle less than a threshold to the vertical direction are selected as valid cracks.

2. The method according to claim 1, characterized in that, In step (1), the image recognition module is implemented using the YOLOv8n model.

3. The method according to claim 2, characterized in that, Step (1) specifically includes: Construct a radar slice image dataset and label valid crack samples using a cross-validation mechanism; The YOLOv8n model was used to train the dataset and identify cracks.

4. The method according to claim 1, characterized in that, In step (3), the time-frequency classification model is the LightGBM model, which is used to classify the A-scan time-frequency features of cracks and normal road surfaces.

5. A ground-penetrating radar-based asphalt pavement base course crack identification device for performing the method according to any one of claims 1-4, characterized in that, include: Image recognition module: used to detect cracks in radar slices and obtain initial crack region bounding boxes; Signal extraction module: used to extract the A-scan signal of the corresponding area in the raw ground penetrating radar data based on the location of the marked box; Time-frequency classification module: Classifies the time-frequency features of the A-scan signal using a time-frequency classification model and outputs crack prediction results; Clustering and filtering module: Performs multi-segment clustering and principal component analysis on the crack prediction results to filter out falsely detected cracks.

6. The apparatus according to claim 5, characterized in that, The image recognition module is implemented using the YOLOv8n model.

7. The apparatus according to claim 5, characterized in that, The time-frequency classification module is implemented using the LightGBM model.

Images