Asphalt pavement base crack identification method and device based on ground penetrating radar

Abstract

The invention discloses an asphalt pavement base crack identification method and device based on a ground penetrating radar, and belongs to the technical field of road engineering pavement maintenance. The method comprises the following steps: 1) carrying out crack image identification on a radar slice map through a YOLOv8n model; 2) based on an identification result, extracting A-scan signals of a crack area, and classifying time-frequency features by using a Light GBM model; and 3) clustering classification results through a'multi-segment clustering + PCA 'analysis method, and filtering false detection cracks. The device is used for executing the method. According to the method, the map and the time-frequency characteristics are fused, the problem that the false detection rate is high due to subjective labeling or objective interference of a single identification method is solved, experiments show that the false detection rate is reduced to 2.8% from 11.2%, and the accuracy and engineering applicability of crack identification are improved.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of pavement maintenance of road engineering, and particularly relates to a method and device for identifying cracks in the base layer of asphalt pavement based on ground penetrating radar. BACKGROUND

[0002] With the development of road maintenance technology, the method for identifying internal pavement diseases based on ground penetrating radar (GPR) has gradually become a research hotspot. The existing technology mainly includes two types: 1. Image recognition method based on deep learning: Through a convolutional neural network (CNN), cracks in radar slice images are automatically detected, replacing traditional manual interpretation, and solving the problems of low efficiency and strong subjectivity of manual processing. However, the performance of this model is affected by the subjectivity of manual annotation, and there are differences in the judgment standards of different annotators for cracks; the limited size of the data set leads to insufficient generalization ability of the model, and the misdiagnosis rate is high in actual application (for example, the actual misdiagnosis rate is 11.2%).

[0003] 2. Recognition method based on time-frequency characteristics of ground penetrating radar: The physical characteristics (such as electromagnetic wave reflection characteristics) of cracks are mined by using time-frequency analysis (such as wavelet transform), reducing the dependence on image annotation. However, electromagnetic signals are easily disturbed by objective factors such as material performance, interlayer adhesion condition and construction quality, resulting in high sensitivity of the model; in complex road environment, the stability is insufficient, and it is difficult to be directly applied to actual engineering.

[0004] The above two methods are complementary in recognition mechanism, but are limited by a single data dimension: image recognition method depends on visual features, which is easily affected by annotation subjectivity; time-frequency analysis method depends on physical signals, which is easily affected by environmental objective factors.

[0005] This leads to high misdiagnosis rate and insufficient reliability of existing GPR data automatic recognition technology in actual road crack detection, which is difficult to meet the engineering requirements. SUMMARY

[0006] The purpose of the embodiment of the present application is to provide a method and device for identifying cracks in the base layer of asphalt pavement based on ground penetrating radar, which combines image recognition results with time-frequency feature analysis, helps to reduce the interference of human subjectivity on recognition accuracy, realizes effective filtering of crack image recognition results, and reduces the crack misdiagnosis rate, so as to solve at least one technical problem involved in the background technology.

[0007] In order to solve the above technical problems, the present application is implemented as follows: The embodiment of the present application provides a method for identifying cracks in the base layer of asphalt pavement based on ground penetrating radar, comprising the following steps: (1) detecting cracks in the radar slice image through an image recognition module to obtain an initial crack region bounding box; (2) extracting A-scan signals in the corresponding region of the ground penetrating radar original data based on the position of the bounding box in step (1); (3) classifying the time-frequency features of the A-scan signals using a time-frequency classification model to output a crack prediction result; (4) performing multi-segment clustering and principal component analysis (PCA) on the crack prediction result to filter false detection cracks, the principal component analysis comprising: extracting marker point coordinates in the crack prediction result; performing multi-segment clustering on the marker points using a DBSCAN algorithm; performing principal component analysis on each clustering result to generate crack fitting line segments; retaining line segments in the vertical direction based on angle screening.

[0008] Optionally, in step (1), the image recognition module is implemented using a YOLOv8n model.

[0009] Optionally, step (1) specifically comprises: constructing a radar slice image dataset and labeling effective crack samples through a cross-verification mechanism; training the dataset and identifying cracks using a YOLOv8n model.

[0010] Optionally, in step (3), the time-frequency classification model is a LightGBM model used to classify the A-scan time-frequency features of cracks and normal road surfaces.

[0011] Optionally, in step (4), the multi-segment clustering and principal component analysis further comprise: converting the radar slice image to HSV color space and extracting crack prediction marker point coordinates; directionally modeling the clustering point set through principal component analysis to generate fitting line segments distributed along the principal axes; screening line segments with an included angle less than a threshold with the vertical direction as effective cracks.

[0012] The application also provides a ground penetrating radar-based asphalt pavement base crack identification device for executing the method, comprising: an image recognition module for detecting cracks in the radar slice image to obtain an initial crack region bounding box; a signal extraction module for extracting A-scan signals in the corresponding region of the ground penetrating radar original data based on the position of the bounding box; a time-frequency classification module for classifying the time-frequency features of the A-scan signals using a time-frequency classification model to output a crack prediction result; The cluster filtering module: multi-section clustering and principal component analysis are performed on the crack prediction result, and false detection cracks are filtered.

[0013] Optionally, the image recognition module is implemented by using a YOLOv8n model.

[0014] Optionally, the time-frequency classification module is implemented by using a LightGBM model.

[0015] Compared with the prior art, the present application has the following beneficial effects: 1. The present application reduces the crack false detection rate from 11.2% of a single image recognition method to 2.8% by combining the dual verification mechanism of ground penetrating radar spectrum recognition and time-frequency feature analysis with the cluster filtering method of “multi-section clustering + PCA”, thereby significantly improving the reliability of the recognition result.

[0016] 2. The present application uses a YOLOv8n model to perform crack preliminary screening on the radar slice image, and the mAP50 reaches 97.4% and the map50-90 reaches 64.6%, which can well balance the crack recognition accuracy and speed; based on the LightGBM model, the crack time-frequency feature is classified, and the accuracy reaches 92.5%, effectively mining the deep physical characteristics of the crack.

[0017] 3. The present application reduces the subjective influence of the data set by using the cross verification labeling mechanism (three independent technical personnel labeling + consistent sample screening); the complementary advantages of the spectrum (anti-objective interference) and the time-frequency (anti-subjective interference) are fused, and the limitations of a single method in complex road conditions are overcome.

[0018] 4. The present application proposes an integrated framework of “image recognition-time-frequency classification-cluster filtering”, realizes the full-process automation from original data analysis, feature extraction to result visualization; designs a multi-channel radar space mapping method (HSV color segmentation + hierarchical dynamic division), accurately associates the crack position with the time-frequency signal; adopts the DBSCAN+PCA clustering algorithm, which is suitable for the tortuous distribution characteristics of the basic crack, accurately fits the crack shape and filters the discrete false detection points.

[0019] 5. The method of the present application fuses “spectrum + time-frequency” to filter 6 false detections of cracks, and 2 cracks are not effectively filtered, and the false detection rate is only 2.8%, which is reduced by 8.4%, verifying the engineering application potential of the method in the road maintenance scene. BRIEF DESCRIPTION OF DRAWINGS

[0020] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other drawings can be obtained from these drawings without creative labor. Figure 1 The flow chart of the asphalt pavement base crack identification method based on ground penetrating radar provided by the embodiments of the present application is shown in Figure 2 The hardware structure schematic diagram of the electronic device provided by the embodiments of the present application is shown in Figure 3 The hardware structure schematic diagram of the electronic device provided by the embodiments of the present application is shown in DETAILED DESCRIPTION

[0021] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the protection scope of the present application.

[0022] The terms "first", "second" and the like in the specification and claims of the present application are used to distinguish similar objects, not to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged under appropriate circumstances, so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second" and the like are generally of a kind, not limited to the number of objects, for example, the first object can be one or more. In addition, "and / or" in the specification and claims indicates at least one of the connected objects, and the character " / ", generally indicates that the front and rear associated objects are in an "or" relationship.

[0023] Please refer to Figure 1 The embodiments of the present application provide an asphalt pavement base crack identification method based on ground penetrating radar, which comprises the following steps: Step S1, detecting cracks in the radar slice image through an image recognition module to obtain an initial crack region annotation box; Step S2, extracting A-scan signals of the corresponding region in the ground penetrating radar original data based on the annotation box position of step S1; Step S3, classifying the time-frequency features of the A-scan signals by using a time-frequency classification model to output a crack prediction result; Step S4, multi-section clustering and principal component analysis are performed on the crack prediction result, and the crack prediction result is filtered, and the principal component analysis comprises: Extracting the marker point coordinates in the crack prediction result; Multi-section clustering is performed on the marker points by using a DBSCAN algorithm; Performing principal component analysis on each clustering result to generate a crack fitting line segment; Reserving the vertical direction crack line segment based on an angle screening.

[0024] In step S1, the image recognition module is realized by using a YOLOv8n model, the map50 of the YOLOv8n model is 97.4%, and the map50-90 is 64.6%, so that the crack recognition accuracy and speed can be well balanced.

[0025] Step S1 specifically comprises: Step S11, constructing a radar slice image data set, and labeling effective crack samples through a cross-checking mechanism; Step S12, training the data set and recognizing cracks by using a YOLOv8n model.

[0026] In step S11, the cross-checking mechanism refers to organizing three technical personnel with radar map interpretation experience, independently labeling the crack slice image, and screening the cracks determined by the three labelers as effective crack samples. The radar slice image is exported as 640 640 pixel points (300 data), and the radar slice image is labeled by using Labelimg, so as to reduce the subjective influence in the data set labeling process.

[0027] In step S3, the time-frequency classification model is a LightGBM model, which is used for classifying the A-scan time-frequency features of cracks and normal road surfaces.

[0028] The LightGBM model is used as the time-frequency classification model in the application, so that the crack classification accuracy can reach 92.5%.

[0029] In step S4, the multi-section clustering and principal component analysis further comprise: Converting the radar slice image to an HSV color space, and extracting crack prediction marker point coordinates; Modeling the direction of the clustering point set by principal component analysis to generate a fitting line segment along the principal axis; Screening the line segment with an included angle less than a threshold value with the vertical direction as an effective crack.

[0030] Experimental verification The method provided by the application and the existing radar slice map method are used to identify 77 cracks, and the experimental results show that the radar slice map method misidentifies 8 crack diseases, and the misidentification rate is 11.2%; and the method of fusing the atlas and the time-frequency in the application filters 6 crack misidentifications, and 2 cracks are not effectively filtered, and the misidentification rate is only 2.8%, and the misidentification rate is reduced by 8.4%. That is, the crack identification method based on the ground penetrating radar "atlas + time-frequency" of the application shows good accuracy and engineering adaptability in the identification of base crack diseases, and verifies the application potential of the method in actual road maintenance.

[0031] The application further provides a ground penetrating radar-based asphalt pavement base crack identification device for performing the method, comprising an image recognition module, a signal extraction module, a time-frequency classification module and a clustering filtering module.

[0032] The image recognition module is used for crack detection on the radar slice map to obtain an initial crack region annotation box. The image recognition module is implemented by using a YOLOv8n model.

[0033] The signal extraction module is used for extracting A-scan signals of the corresponding region in the ground penetrating radar original data based on the annotation box position.

[0034] The time-frequency classification module classifies the time-frequency features of the A-scan signals by using a time-frequency classification model, and outputs a crack prediction result. The time-frequency classification module is implemented by using a LightGBM model.

[0035] The clustering filtering module performs multi-segment clustering and principal component analysis on the crack prediction result to filter misidentified cracks.

[0036] Referring to Figure 2 The embodiment of the application further provides an electronic device 700, which comprises a processor 710, a memory 709, a program or instruction stored in the memory 709 and executable on the processor 710, and the program or instruction is executed by the processor 710 to realize each process of the above-mentioned ground penetrating radar-based asphalt pavement base crack identification method embodiment, and the same technical effect can be achieved. To avoid repetition, it will not be repeated here.

[0037] It should be noted that the electronic device in the embodiment of the application includes the mobile electronic device and the non-mobile electronic device described above.

[0038] Figure 3 A hardware structure schematic diagram of an electronic device for implementing the embodiment of the application.

[0039] The electronic device 700 includes, but is not limited to, a radio frequency unit 701, a network module 702, an audio output unit 703, an input unit 704, a sensor 705, a display unit 706, a user input unit 707, an interface unit 708, a memory 709, and a processor 710, etc.

[0040] Those skilled in the art can understand that the electronic device 700 can further include a power supply (such as a battery) for supplying power to each component, and the power supply can be logically connected to the processor 710 through a power management system, so as to realize functions such as management of charging, discharging, and power consumption management through the power management system. Figure 3 The electronic device structure shown in the figure is not a limitation on the electronic device, and the electronic device can include more or fewer components than shown, or combine certain components, or different component arrangements, which are not described here.

[0041] It should be understood that in the embodiments of the present application, the input unit 704 can include a graphics processor (GPU) 7041 and a microphone 7042, and the graphics processor 7041 processes image data of a still image or a video obtained by an image capture device (such as a camera) in a video capture mode or an image capture mode. The display unit 706 can include a display panel 7061, which can 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 can include two parts of a touch detection device and a touch controller. The other input devices 7072 can include, but are not limited to, a physical keyboard, function keys (such as volume control keys, on-off keys, etc.), trackballs, mice, joysticks, etc., which are not described here. The memory 709 can be used to store software programs and various data, including but not limited to application programs and operating systems. The processor 710 can integrate an application processor and a modem processor, wherein the application processor mainly processes operating systems, user interfaces, and application programs, etc., and the modem processor mainly processes wireless communication. It can be understood that the above-mentioned modem processor can also not be integrated into the processor 710.

[0042] The embodiments of the present application also provide a readable storage medium, the readable storage medium stores a program or instructions, the program or instructions are executed by a processor to realize each process of the above-mentioned ground penetrating radar based asphalt pavement base crack identification method embodiment, and can achieve the same technical effect, to avoid repetition, which is not described here.

[0043] The processor is the processor in the electronic device described in the above embodiments. The readable storage medium includes a computer readable storage medium, such as a computer read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, etc.

[0044] The embodiment of the present application further provides a chip, which comprises a processor and a communication interface, the communication interface is coupled with the processor, the processor is used for running programs or instructions to realize each process of the above-mentioned asphalt pavement base crack identification method based on a ground penetrating radar and achieve the same technical effect, and details are not repeated here to avoid repetition.

[0045] It should be understood that the chip mentioned in the embodiment of the present application can also be referred to as a system-level chip, a system chip, a chip system or a system-on-chip chip, etc.

[0046] It should be noted that in this paper, the term "includes", "contains" or any other variant thereof is intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed, or includes elements inherent to such process, method, article or device. Without more limitations, the element defined by the statement "includes a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

[0047] In addition, it should be noted that the scope of the method and system in the embodiment of the present application is not limited to performing functions in the order shown or discussed, but can also include performing functions in a substantially simultaneous manner or in the opposite order according to the functions involved, for example, the described method can be performed in an order different from that described, and various steps can also be added, omitted or combined. In addition, the features described with reference to certain examples can be combined in other examples.

[0048] The embodiments of the present application are described above in combination with the drawings, but the present application is not limited to the above-mentioned specific embodiments, and the above-mentioned specific embodiments are only illustrative, not restrictive, and those skilled in the art can make many forms under the inspiration of the present application without departing from the scope of the present application and the scope protected by the claims, which all belong to the protection of the present application.

Claims

1. A method for identifying cracks in asphalt pavement base based on ground penetrating radar, characterized in that: The following steps are involved: (1) Use the image recognition module to detect cracks in the radar slice image and obtain the initial crack area annotation box; (2) Based on the position of the annotation box in step (1), extract the A-scan signal of the corresponding area in the ground penetrating radar raw data; (3) Using a time-frequency classification model to classify the time-frequency features of the A-scan signal and output a crack prediction result; (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 marked points in the crack prediction results; The DBSCAN algorithm is used to perform multi-segment clustering of the marker points; Perform principal component analysis on each clustering result to generate crack fitting segments; Perpendicular crack segments are retained based on angle screening.

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 annotate valid crack samples through a cross-verification mechanism; The YOLOv8n model is used to train the dataset and identify cracks.

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

5. The method according to claim 1, wherein In step (4), multi-segment clustering and principal component analysis further includes: Convert the radar slice image to HSV color space and extract the coordinates of the crack prediction marker points; The cluster point set is oriented modeled by principal component analysis to generate fitting line segments distributed along the principal axis; The line segments whose angle with the vertical direction is less than the threshold are selected as effective cracks.

6. A device for identifying cracks in asphalt pavement base based on ground penetrating radar for executing the method according to any one of claims 1 to 5, characterized in that: include: Image recognition module: used to detect cracks on radar slice images and obtain the initial crack area annotation box; Signal extraction module: used to extract the A-scan signal of the corresponding area in the ground penetrating radar raw data based on the position of the annotation 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.

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

8. The device according to claim 6, characterized in that The time-frequency classification module is implemented using the LightGBM model.

Images