Crack detection method and system for expressway construction

Abstract

The invention belongs to the technical field of construction crack detection, and particularly discloses a crack detection method and system for expressway construction, and the method comprises the steps: synchronously collecting multiband data through a multispectral thermal imaging sensor, and integrating the surface morphology, material reflection and deep thermal radiation characteristics through a multisource registration fusion technology; the crack identification capability in a complex environment is enhanced, the crack edge continuity and thermal radiation fluctuation characteristics are further analyzed through dynamic correlation, transient interference is accurately stripped, the problem of high false alarm is solved, meanwhile, a scoring model is constructed in combination with geometric and thermodynamic parameters, the damage degree and the expansion trend are quantified, and the accuracy of crack identification is improved. And finally, the crack expansion risk is pre-judged through the dynamic change trend of the thermodynamic parameters, a hierarchical early warning signal is triggered to form a sensing-anti-interference-assessment-decision closed loop chain, the whole process optimization from data fusion and interference suppression to risk assessment and maintenance response is realized, and the accuracy of road crack detection is guaranteed.

Description

Technical Field

[0001] The present invention belongs to the technical field of construction crack detection, and specifically relates to a crack detection method and system for highway construction. Background Art

[0002] Cracks are a common form of early diseases on highways and may be caused by construction defects, material aging, overloading, or environmental factors. If not detected and treated in time, cracks may expand and cause subgrade damage and road surface collapse, threatening driving safety.

[0003] The prior art, such as a subgrade crack detection, identification and repair method based on infrared thermal imaging analysis disclosed in a Chinese invention patent application with the application number CN202110551750.8, fixes an infrared thermal imaging device and a visible light imaging device by laying rails, synchronously collects infrared thermal images and visible light images of the road surface, manually processes the infrared data to generate a temperature contour map, combines the visible light images to identify hidden cracks, and predicts the crack development trend based on the temperature distribution.

[0004] Another prior art, such as an asphalt pavement crack development degree detection method based on infrared thermal image analysis disclosed in a Chinese invention patent application with the application number CN201680089245.3, establishes a crack development degree model, obtains the measured temperature difference data between the crack area and the road surface by using infrared thermal images, combines the atmospheric temperature correction model to calculate the crack development degree, and simultaneously fuses the gray scale information and the temperature information for detection.

[0005] For asphalt pavements, the core of the Chinese invention patent application with the application number CN202110551750.8 is the dual-modal data fusion of infrared and visible light and manual-assisted analysis, which is suitable for small-scale static detection but is restricted by rails and lacks flexibility. The core of the Chinese invention patent application with the application number CN201680089245.3 is the static temperature difference model and atmospheric temperature compensation, but it depends on a preset model. Combining these two technical solutions, both are based on the fusion of infrared thermal imaging and visible light images, and identify cracks through temperature difference or temperature distribution. There are still the following deficiencies: 1. The existing solutions only fuse infrared thermal imaging and visible light images, and do not make full use of multi-spectral or thermodynamic dynamic parameters, and the data fusion dimension is still insufficient.

[0006] 2. Dependent on static temperature thresholds or manual correction, unable to respond to dynamic interferences such as steam and dust in real time, and has a high false alarm rate in the scenario of steam and oil stains mixed after heavy rain.

[0007] 3. The prior art only qualitatively evaluates cracks through temperature difference or gray scale information, lacks quantification of risks, and is not convenient for subsequent processing. Summary of the Invention

[0008] In view of this, a crack detection method and system for highway construction are proposed to solve the limitations of the existing technology in terms of poor environmental adaptability, sensitivity to optical pollution, and rough feature analysis in moisture and false alarm prevention.

[0009] The object of the present invention can be achieved by the following technical solutions: In the first aspect of the present invention, a crack detection method for highway construction is provided, and the method includes the following steps: S1. Synchronously collect multi-band image data on the surface of the asphalt layer through a multi-spectral thermal imaging sensor, and perform registration and fusion, where the multi-bands include a visible light band, a near-infrared band, and a long-wave infrared band.

[0010] S2. Eliminate false crack interference data based on the correlation analysis of edge continuity features and thermal radiation fluctuation features, and obtain effective multi-band image data.

[0011] S3. Extract the geometric shape parameters and thermodynamic parameters of the cracks from the effective multi-band image data, and calculate the current comprehensive risk coefficient through a weighted scoring model.

[0012] S4. Generate a hierarchical thermal map based on the spatial distribution of the current comprehensive risk coefficient, and trigger a hierarchical warning signal in combination with the expansion trend to generate a crack detection report.

[0013] In the second aspect of the present invention, a crack detection system for highway construction is provided, and the system includes the following modules: A multi-spectral thermal imaging acquisition module that synchronously collects multi-band image data on the surface of the asphalt layer through a multi-spectral thermal imaging sensor and performs registration and fusion.

[0014] A crack interference identification and elimination module that eliminates false crack interference data based on the correlation analysis of edge continuity features and thermal radiation fluctuation features, and obtains effective multi-band image data.

[0015] A crack risk analysis module that extracts the geometric shape parameters and thermodynamic parameters of the cracks from the effective multi-band image data, and calculates the current comprehensive risk coefficient through a weighted scoring model.

[0016] A crack report generation terminal that generates a hierarchical thermal map based on the spatial distribution of the current comprehensive risk coefficient, and triggers a hierarchical warning signal in combination with the expansion trend to generate a crack detection report.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) The present invention synchronously collects multi-band data through a multi-spectral thermal imaging sensor, and uses multi-source registration and fusion technology to integrate surface morphology, material reflection, and deep thermal radiation characteristics, enhancing the ability to identify cracks in complex environments. By dynamically correlating and analyzing the edge continuity of cracks and the fluctuation characteristics of thermal radiation, transient interference is accurately stripped, solving the problem of high false alarms. A scoring model is constructed by combining geometric and thermodynamic parameters to quantify the damage degree and expansion trend, generating a thermal map and predicting risk triggering warnings, realizing the optimization of the entire process, and improving the detection accuracy and environmental adaptability.

[0018] (2) The present invention overcomes the limitations of a single sensor in complex environments by performing multi-band synchronous acquisitions such as visible light, near-infrared, and long-wave infrared, significantly enhancing the contrast between cracks and the background, effectively solving the problems of missed detection and false detection caused by insufficient data dimensions in traditional methods, and providing a high-precision, multi-dimensional data basis for subsequent dynamic interference rejection and quantitative evaluation.

[0019] (3) The present invention performs dynamic interference rejection through the correlation analysis of edge continuity features and thermal radiation fluctuation features, breaking through the limitations of traditional static thresholds or manual corrections, realizing real-time discrimination and rejection of interference, facilitating accurate distinction between real cracks and transient interferences such as steam and dust, reducing the false alarm rate, and ensuring the reliability of subsequent crack detection.

[0020] (4) The present invention constructs a weighted scoring model by combining the geometric shape of cracks and thermodynamic dynamic parameters to calculate the current comprehensive risk coefficient, realizing the quantitative risk assessment of cracks, making up for the defect of relying solely on qualitative assessment of temperature difference. At the same time, by generating a spatial distribution thermal map and a crack detection report, the risk level distribution is intuitively quantified, significantly reducing the engineering safety hazards caused by evaluation deviations. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for describing the embodiments will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0022] Figure 1 It is a schematic flow chart of the implementation steps of the method of the present invention.

[0023] Figure 2 It is a schematic connection diagram of the system module structure of the present invention.

[0024] Figure 3 It is a schematic flow chart of the false crack interference data rejection of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0025] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0026] Please refer to Figure 1 As shown, the present invention provides a crack detection method for highway construction, and the method includes: S1. Synchronously collect multi-band image data on the surface of the asphalt layer through a multi-spectral thermal imaging sensor, and perform registration and fusion, where the multi-bands include a visible light band, a near-infrared band, and a long-wave infrared band.

[0027] Specifically, the specific execution process of the S1 step includes: S11. Synchronously collect images on the surface of the asphalt layer through a multi-spectral sensor in the visible light band, the near-infrared band, and the long-wave infrared band.

[0028] S12. Use an image registration algorithm to perform spatial alignment and fusion on the multi-band images to generate a fused image containing temperature gradient distribution and texture features.

[0029] S13. Mark the geographical location coordinates for each acquisition area, and bind the fused image with the coordinate information.

[0030] It should be added that the registration and fusion are performed using an image registration algorithm, where the image registration algorithm is a relatively mature existing algorithm, and its specific execution process will not be elaborated here.

[0031] By synchronously collecting in multiple bands such as visible light, near-infrared, and long-wave infrared, the embodiments of the present invention overcome the limitations of a single sensor in a complex environment, significantly improve the contrast between cracks and the background, and effectively solve the problems of missed detection and false detection caused by insufficient data dimensions in traditional methods, providing a high-precision and multi-dimensional data basis for subsequent dynamic interference elimination and quantitative evaluation.

[0032] Please refer to Figure 3 As shown, S2. Eliminate false crack interference data based on the correlation analysis of edge continuity features and thermal radiation fluctuation features, and obtain effective multi-band image data.

[0033] Specifically, the specific elimination process of eliminating false crack interference data includes: S21. Perform grayscale processing on the visible light band image, set the sliding window size based on the minimum apparent width of the crack, traverse and calculate the texture roughness of each pixel neighborhood, and mark the area with texture roughness lower than the preset threshold interval as the low roughness area.

[0034] S22. Extract the time-series fluctuation data of the thermal radiation intensity in the low-roughness region from the thermal radiation image sequence collected synchronously in the near-infrared band, calculate its standard deviation and mean offset, and simultaneously calculate the deviation value between the temperature distribution in the long-wave infrared band image and the corresponding brightness distribution in the visible light image.

[0035] S23. Extract the edge continuity features of the detection region from the visible light band image, combine with the thermal radiation intensity fluctuation features in the near-infrared band, construct a texture-thermal radiation correlation matrix, and calculate the correlation coefficient between the two.

[0036] S24. Perform interference determination on the standard deviation, mean offset, deviation value, and correlation degree through a pre-set interference determination rule. For the regions determined to be interfered, synchronously delete the image data in the visible light, near-infrared, and long-wave infrared bands.

[0037] Further, the size setting of the sliding window is associated with the minimum apparent width of the target crack, specifically satisfying: , is the width of the sliding window, represents the minimum apparent width of the target crack, and the traversal step of the sliding window is set to 1 / 2 of the window width to ensure that adjacent windows have a 50% overlapping area.

[0038] It should be added that the preset threshold interval of texture roughness can be determined by statistically analyzing the roughness distribution of historical crack samples.

[0039] Further, the mean offset is the absolute deviation percentage of the average value of the time-series data of the thermal radiation intensity within the observation period relative to the thermal radiation intensity value of the surrounding area.

[0040] It should be added that real cracks may have large thermal radiation changes due to different materials or structural problems, while false cracks such as stains and shadows may have smaller fluctuations or different patterns. However, a threshold needs to be set, and being too high or too low may cause problems. Similarly, real cracks may have different thermal characteristics, resulting in a large mean offset.

[0041] It can be understood that due to the irregularity of the internal structure and the complexity of the heat conduction path of real cracks, their thermal radiation fluctuations usually have a high standard deviation. While surface attachments such as paint and dust have more uniform thermal radiation fluctuations due to the homogeneity of the material, and the standard deviation is significantly lower than that of real cracks. At the same time, the difference in thermal inertia between the surface attachment and the substrate causes an obvious mean offset in the time-series data of thermal radiation, such as different heat absorption / dissipation rates of the attachment. Therefore, analysis is carried out from two parameter indicators of the standard deviation and the mean offset.

[0042] Further, the deviation value between the temperature distribution and the brightness distribution is calculated by the formula where, represents the deviation value between the temperature distribution and the brightness distribution is the temperature standard deviation of the shadow area in the long-wave infrared image is the brightness standard deviation of the corresponding area in the visible light image is the band response coefficient determined through calibration experiments, with the unit of ℃ / gray value. Its physical meaning is the temperature standard deviation corresponding to the unit brightness standard deviation

[0043] Among them, the calibration experiment refers to collecting multiple groups of shadow-free samples under a standard light environment, calculating their temperature and brightness standard deviations respectively, and fitting through the least squares method the value so that and have the same unit

[0044] Furthermore, the correlation coefficient can be calculated through the Pearson correlation coefficient formula after normalizing the edge continuity feature vector and the thermal radiation fluctuation data sequence. Its specific calculation formula is: , represents the correlation coefficient is the edge continuity feature is the thermal radiation fluctuation feature is the covariance operation and respectively represent the standard deviations of the edge continuity feature and the thermal radiation fluctuation feature

[0045] It can be understood that the Pearson correlation coefficient is standardized by dividing the covariance by the product of the standard deviations, eliminating the influence of dimensions, making the results more comparable. At the same time, it also ensures that changes in the position and scale of the variables do not affect the coefficient, and can accurately reflect the strength and direction of the linear relationship between the edge continuity feature and the thermal radiation fluctuation feature, providing a reliable quantitative basis for analyzing the correlation between the two, and facilitating an intuitive judgment of the strength and direction of the correlation

[0046] Furthermore, the specific determination process for interference determination includes: if the standard deviation of the thermal radiation intensity is lower than the lower limit of the standard deviation reference range of the crack feature model and its mean offset exceeds the preset offset threshold, it is determined that there is interference from surface attachments

[0047] If the deviation value exceeds the preset adjustment threshold, it is determined that there is interference from optical shadow artifacts

[0048] If the correlation coefficient is lower than the preset correlation coefficient threshold, it is determined that there is interference from stains or foreign objects

[0049] In a specific embodiment, the thermal radiation standard deviation reference range, the preset correlation coefficient threshold, and the preset adjustment threshold are generated through joint calibration of multi-modal crack data and are dynamically updated in the crack feature model. Exemplarily, the thermal radiation standard deviation reference range is determined through the following steps: (a) Collect the thermal radiation time-series fluctuation data of the known true crack area.

[0050] (b) Calculate the standard deviation of each true crack area and generate a standard deviation distribution histogram.

[0051] (c) Use the 5th percentile and the 95th percentile of the distribution histogram as the lower limit and the upper limit of the reference range respectively to form the standard deviation reference range.

[0052] It should be noted that by taking the interval from the 5th to the 95th percentile, extreme standard deviations that are too high or too low can be excluded, and the data range consistent with the thermal radiation fluctuation characteristics of true cracks can be retained. The low standard deviation below the 5th percentile may correspond to surface attachments such as paint and stains, while the high standard deviation above the 95th percentile may be caused by non-crack heat sources such as local thermal radiation anomalies, and both are excluded. And this percentile interval can be dynamically adjusted based on the update of the true crack sample library.

[0053] In the embodiment of the present invention, dynamic interference rejection is performed through the correlation analysis of the edge continuity feature and the thermal radiation fluctuation feature, breaking through the limitations of traditional static thresholds or manual corrections, realizing real-time discrimination and rejection of interference, facilitating accurate distinction between true cracks and transient interferences such as steam and dust, reducing the false alarm rate, and ensuring the reliability of subsequent crack detection.

[0054] S3. Extract the geometric shape parameters and thermodynamic parameters of the crack from the effective multi-band image data, and calculate the current comprehensive hazard coefficient through a weighted scoring model.

[0055] Specifically, the geometric shape parameters of the crack include crack length, average crack width, crack horizontal development change rate, crack branch density, and the change gradient of crack width with depth. The specific extraction process includes: A1. Use the Canny edge detection algorithm to identify the center of the crack in the visible light band image, measure the longest continuous pixel distance along the crack center line, and convert it into the actual length in combination with the image resolution as the crack length.

[0056] A2. Perform multi-point sampling of the width in the direction perpendicular to the crack center line and calculate the average value as the average crack width.

[0057] A3. Measure the width in segments along the crack development direction, and fit the slope of the change in width with the extension distance through linear regression as the crack width horizontal development change rate.

[0058] A4. Count the number of crack bifurcation points from the visible light band image, and calculate the crack branch density by comparing it with the area within the crack region.

[0059] A5. Extract the cross-section profile corresponding to the crack in the visible light band image, and then measure the width layer by layer along the depth direction to calculate its change gradient with depth.

[0060] Understandably, the Canny edge detection algorithm is a relatively mature existing algorithm, and its specific detection process and recognition principle will not be elaborated here.

[0061] It should be added that the layer spacing along the depth direction is dynamically adjusted according to the apparent width of the crack, and the change gradient is quantified by the ratio of the width difference between adjacent layers to the depth increment.

[0062] Along the extension direction of the crack center line, cross-sections are intercepted at equal intervals, and the interval is not greater than 2 times the apparent width of the crack.

[0063] Exemplarily, the dynamic interval stratification can be divided according to the following rules: surface layer interval, depth is 0 - 5 mm, layer spacing is 0.5 mm. Middle layer interval, depth is 5 - 20 mm, layer spacing is 1 mm. Deep layer interval, depth > 20 mm and layer spacing is 2 mm.

[0064] Furthermore, the specific calculation process of the change gradient includes: B1. Intercept cross-sections at equal intervals along the extension direction of the crack center line, and stratify them at dynamic intervals along the depth direction on each cross-section.

[0065] B2. Measure the maximum and minimum widths within each layer, and calculate the width volatility within the layer.

[0066] B3. If the width volatility within the layer is less than the set threshold, calculate the average width change gradient of adjacent layers in sequence along the depth direction.

[0067] It should be added that the width volatility within the layer can help distinguish real crack propagation and external interference. For example, if the width change of a certain layer is large, it may mean that there are foreign objects or measurement errors in that layer, rather than real crack changes. By calculating the volatility, these unreliable data points can be identified and processed to improve the overall detection accuracy.

[0068] Understandably, the width volatility within the layer is obtained by comparing the difference between the maximum and minimum measured widths within the layer with the average width within the layer.

[0069] It should be added that the specific calculation formula of the change gradient is: , where represents the th depth layer and the Gradient of average width change of each depth layer, represents the depth layer number, , and respectively represent the average widths of the -th depth layer and the -th depth layer, respectively represent the depth coordinates of the -th depth layer and the -th depth layer.

[0070] It should also be added that if the width volatility within the layer is greater than or equal to the set threshold, it is determined as foreign object occlusion, such as gravel stuck in a crack, and an error correction mechanism is required. Specifically, the error correction mechanism includes: performing sub-layer analysis and scanning on this layer, compressing the layer interval to 1 / 5 of the original setting, using Gaussian filtering to smooth the width data, and recalculating the average width change gradient of adjacent layers after removing outliers.

[0071] Exemplarily, the sub-layer scanning compresses the 0.5mm interval to 0.1mm to identify the boundary of tiny foreign objects, and the Gaussian filtering eliminates high-frequency noise through the filter and retains the true gradient trend.

[0072] Specifically, the thermodynamic parameters include the lateral temperature gradient, the longitudinal temperature decay rate, and the heat diffusion area. The specific extraction process includes: R1. Using the Canny edge detection algorithm to identify the crack edge and extract the continuous contour.

[0073] R2. Selecting temperature measurement points at equal intervals on both sides of the contour in the long-wave infrared image, and calculating the ratio of the average temperature difference between the two sides to the actual width of the crack to obtain the lateral temperature gradient.

[0074] R3. Dividing the temperature measurement segments at equal intervals along the contour extension direction, and performing linear fitting on the temperature data of each segment to obtain the longitudinal temperature decay rate.

[0075] R4. Extracting the extended area outside the crack contour based on the pre-set temperature boundary, and taking the area of the extended area outside the crack contour as the heat diffusion area.

[0076] In a specific embodiment, the extended area outside the crack contour can be the boundary where the crack edge extends outward until the temperature decays to the ambient temperature ±2°C. The crack contour is expanded outward through the morphological dilation algorithm until the temperature field meets , and the pixel area covered by this extended area is the heat diffusion area. Wherein, is the temperature value at the current position , is the ambient temperature value.

[0077] Understandably, the calculation formula for the longitudinal temperature decay rate is: , represents the longitudinal temperature attenuation rate, is expressed as the temperature change amount, represents the length of the temperature measurement section in the extension direction.

[0078] More specifically, the calculation of the comprehensive hazard coefficient includes: E1. Calculate the basic hazard score according to the crack length and the average crack width, and perform non-linear correction on the score through the crack horizontal development change rate, the crack branch density, and the change gradient of the crack width with depth to obtain the corrected basic hazard score.

[0079] It should be added that calculating the basic hazard score according to the crack length and the average crack width includes: normalizing the crack length and the average crack width, and then performing weighted summation calculation to obtain the basic hazard score. Among them, the weights of the crack length and the average crack width are set corresponding to the crack type. For tensile cracks, the weights of the crack length and the average crack width can be taken as 0.6 and 0.4 respectively, and for shear cracks, the weights of the crack length and the average crack width can be taken as 0.4 and 0.6 respectively.

[0080] Understandably, when calculating the basic hazard score, the weight settings of the crack length and the average crack width need to comprehensively consider the influence of both on the basic structure hazard degree. By collecting a large amount of historical data, such as the crack length, the average crack width, and the corresponding actual evaluation results of the basic hazard under different working conditions and environmental conditions, using regression analysis or machine learning algorithms such as linear regression and support vector machine to determine the optimal values of the weights of the crack length and the average crack width, ensuring that the weights can reflect the actual influence of both in different scenarios. After determining the optimal weights, verify the rationality of the scoring formula through simulation or actual cases to ensure that the basic hazard score accurately reflects the actual hazard degree. Substitute the real-time measured crack length and average crack width into the formula to calculate the score, providing a scientific basis for the maintenance and treatment of the basic structure. This weight setting comprehensively considers the contributions of the crack length and the average width to the basic hazard degree, realizing the accurate assessment of the basic hazard.

[0081] It should also be added that the non-linear correction of the score specifically includes: performing maximum-minimum normalization processing on the crack horizontal development change rate, the crack branch density, and the change gradient of the crack width with depth, and respectively recording the processing results as 、 and , calculate the correction coefficient , where the specific calculation formula of the correction coefficient is: , is the natural constant, 、 and They respectively represent the weights corresponding to the set crack horizontal development change rate, crack branch density, and the change gradient of crack width with depth.

[0082] Understandably, in actual engineering, the relationships between indicators such as the crack horizontal development change rate, crack branch density, and the change gradient of crack width with depth and the final score are often non - linear. For example, after a certain index exceeds a certain threshold, its impact on the score may increase significantly, while a linear model cannot capture such changes. Through non - linear functions such as exponential functions, the complex relationship between the index and the score can be more accurately fitted, making the correction coefficient more in line with the actual situation. This formula can comprehensively consider the non - linear correction of multiple factors on the basic risk score, improve the accuracy and reliability of the evaluation results, and provide a more accurate basis for engineering decisions.

[0083] It should also be added that 、 and are respectively used to measure the relative importance of the crack horizontal development change rate, crack branch density, and the change gradient of crack width with depth in correcting the basic risk score. The larger the weight value, the more significant the impact of the corresponding factor on the correction coefficient. For example, when is larger, it indicates that the crack horizontal development change rate plays a more crucial role in correcting the risk score. The setting of these weights needs to be determined by combining historical engineering data and statistical analysis. It can be optimized through regression analysis, least - squares method, etc. Based on the actual risk assessment results, the weights are trained and optimized to make accurately reflect the actual risk level under the combined action of various factors. After determining the weights, it is also necessary to verify through multiple groups of measured data to ensure that under different working conditions, the correction coefficient can reasonably reflect the influence of each factor. For example, when the crack branch density is high and is larger, should increase significantly to reflect the increase in the risk level.

[0084] E2. Calculate the thermodynamic score based on thermodynamic parameters, and fuse it with the corrected basic risk score according to the preset weight to obtain the current comprehensive risk coefficient.

[0085] Specifically, the thermodynamic parameters are processed in the same way as the principle of the correction coefficient to obtain the thermodynamic score.

[0086] Understandably, the weight values of the thermodynamic score and the basic risk score are determined by combining the analytic hierarchy process and data - driven methods such as the entropy weight method. Specifically, a judgment matrix can be constructed through expert experience to quantitatively evaluate the relative importance of geometric morphology and thermodynamic anomalies to structural safety, such as the basic score weight of 0.6 and the thermodynamic score weight of 0.4, and the weight ratio is verified by regression and automatically optimized in combination with historical accident data.

[0087] S4. Generate a spatial distribution grading heat map based on the current comprehensive risk coefficient, trigger a grading warning signal in combination with the expansion trend, and generate a crack detection report.

[0088] It should be added that the surface of the asphalt layer is divided into grids to obtain each collection area. The steps S1 to S3 are repeated for each collection area to obtain the comprehensive risk coefficient of each collection area. Then, the comprehensive risk coefficient of each collection area is mapped to the electronic map of the construction area to generate a grading heat map.

[0089] In the embodiment of the present invention, a weighted scoring model is constructed by combining the crack geometric morphology and thermodynamic dynamic parameters to calculate the current comprehensive risk coefficient, realizing the risk quantitative assessment of cracks, making up for the defect of only relying on qualitative assessment of temperature difference. At the same time, by generating a spatial distribution heat map and a crack detection report, the risk level distribution is visually quantified, significantly reducing the engineering safety hazards caused by assessment deviation.

[0090] Furthermore, triggering the grading warning signal includes: obtaining the matching warning level after matching the comprehensive risk coefficient with the associated comprehensive risk coefficients under each pre-set warning level, and taking it as the initial warning level.

[0091] Obtain the comprehensive risk coefficient sequence of the current and historical detection periods, and calculate the change rate of the risk coefficient between adjacent periods.

[0092] Count the number of detection periods in which the risk coefficient continuously increases in the sequence, and calculate its continuous growth ratio with the total number of detection periods.

[0093] When the change rate of the risk coefficient exceeds the first pre-set warning threshold, or the continuous growth ratio exceeds the second pre-set warning threshold, the initial warning level is raised to the next warning level, triggering the warning signal of the next warning level.

[0094] It should also be added that the crack detection report automatically integrates multi-spectral images, grading heat maps, multi-band image data and grading warning signals.

[0095] In the embodiment of the present invention, a weighted scoring model is constructed by combining the crack geometric morphology and thermodynamic dynamic parameters to calculate the current comprehensive risk coefficient, realizing the risk quantitative assessment of cracks, making up for the defect of only relying on qualitative assessment of temperature difference. At the same time, by generating a spatial distribution heat map and a crack detection report, the risk level distribution is visually quantified, significantly reducing the engineering safety hazards caused by assessment deviation.

[0096] Please refer to Figure 2 As shown, the present invention provides a crack detection system for highway construction, which includes: a multi-spectral thermal imaging acquisition module, a crack interference identification and elimination module, a crack risk analysis module and a crack report generation terminal.

[0097] Among them, the crack interference identification and elimination module is respectively connected to the multispectral thermal imaging acquisition module and the crack risk analysis module, and the crack risk analysis module is connected to the crack report generation terminal.

[0098] The multispectral thermal imaging acquisition module synchronously acquires multi-band image data on the surface of the asphalt layer through a multispectral thermal imaging sensor and performs registration and fusion.

[0099] The crack interference identification and elimination module eliminates false crack interference data based on the correlation analysis of edge continuity features and thermal radiation fluctuation features and obtains effective multi-band image data.

[0100] The crack risk analysis module extracts the geometric shape parameters and thermodynamic parameters of the crack from the effective multi-band image data and calculates the current comprehensive risk coefficient through a weighted scoring model.

[0101] The crack report generation terminal generates a graded heat map according to the spatial distribution of the current comprehensive risk coefficient, triggers a graded early warning signal in combination with the expansion trend, and generates a crack detection report.

[0102] The preset parameters in the above formula are set by those skilled in the art according to the actual situation.

[0103] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product.

[0104] Those of ordinary skill in the art can realize that the modules and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0105] In addition, the functional modules in each embodiment of the present application can be integrated into one processing module, or each module can exist physically alone, or two or more modules can be integrated into one module.

[0106] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person familiar with the technical field of this application can easily think of changes or substitutions within the technical scope disclosed in this application, and all should be covered by the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

[0107] Finally, the above are only the preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present invention shall be included within the protection scope of the present invention.

Claims

1. A crack detection method for highway construction, characterized in that: The method includes the following steps: S1. Synchronously collect multi-band image data on the surface of the asphalt layer through a multi-spectral thermal imaging sensor, and perform registration and fusion. Among them, the multi-bands include the visible light band, the near-infrared band, and the long-wave infrared band; S2. Eliminate false crack interference data based on the correlation analysis of edge continuity features and thermal radiation fluctuation features, and obtain effective multi-band image data; S3. Extract the geometric shape parameters and thermodynamic parameters of the cracks from the effective multi-band image data, and calculate the current comprehensive hazard coefficient through a weighted scoring model; S4. Generate a spatially distributed hierarchical thermal map according to the current comprehensive hazard coefficient, and combine the expansion trend to trigger a hierarchical warning signal to generate a crack detection report.

2. The crack detection method for highway construction according to claim 1, characterized in that: The specific execution process of step S1 includes: Synchronously collect images on the surface of the asphalt layer through a multi-spectral sensor in the visible light band, the near-infrared band, and the long-wave infrared band; Use an image registration algorithm to perform spatial alignment and fusion on the multi-band images to generate a fused image containing temperature gradient distribution and texture features; Mark the geographical location coordinates for each acquisition area, and bind the fused image with the coordinate information.

3. A crack detection method for highway construction according to claim 1, characterized in that: The specific elimination process of eliminating false crack interference data includes: Perform grayscale processing on the visible light band image, set the sliding window size based on the minimum apparent width of the crack, traverse and calculate the texture roughness of each pixel neighborhood, and mark the area with texture roughness lower than the preset threshold range as the low roughness area; In the thermal radiation image sequence synchronously collected in the near-infrared band, extract the temporal fluctuation data of the thermal radiation intensity in the low roughness area, calculate its standard deviation and mean offset, and at the same time calculate the deviation value between the temperature distribution in the shadow area in the long-wave infrared band image and the corresponding brightness distribution in the visible light image; Extract the edge continuity features of the detection area from the visible light band image, combine with the thermal radiation intensity fluctuation features in the near-infrared band, construct a texture-thermal radiation correlation matrix, and calculate the correlation coefficient between the two; Perform interference determination on the standard deviation, mean offset, deviation value, and correlation degree through a preset interference determination rule. For the area determined to be interference, synchronously delete its image data in the visible light, near-infrared, and long-wave infrared bands.

4. A crack detection method for highway construction according to claim 3, characterized in that: The specific determination process of performing interference determination includes: If the standard deviation of the thermal radiation intensity is lower than the lower limit of the standard deviation reference range of the crack feature model and its mean offset exceeds the preset offset threshold, it is determined that there is surface attachment interference; If the deviation value exceeds the preset adjustment threshold, it is determined that there is optical shadow artifact interference; If the correlation coefficient is lower than the preset correlation coefficient threshold, it is determined that there is stain or foreign object interference.

5. A crack detection method for highway construction according to claim 1, characterized in that: The geometric shape parameters of the crack include crack length, average crack width, crack horizontal development change rate, crack branch density, and the change gradient of crack width with depth. The specific extraction process includes: Use the Canny edge detection algorithm to identify the crack center in the visible light band image, measure the longest continuous pixel distance along the crack center line, and convert it into the actual length in combination with the image resolution as the crack length; Perform multi-point sampling widths in the direction perpendicular to the crack center line, and calculate the average value as the average crack width; Measure the widths in segments along the crack development direction, and fit the slope of the change in width with the extension distance through linear regression as the horizontal development change rate of the crack width; Count the number of crack bifurcation points from the visible light band image, and calculate the crack branch density by taking the ratio to the area within the crack region; Extract the cross-sectional profile corresponding to the crack in the visible light band image, and then measure the widths in layers along the depth direction and calculate the change gradient with depth.

6. The crack detection method for highway construction according to claim 5, characterized in that: The specific calculation process of the change gradient includes: Intercept cross-sections at equal intervals along the extension direction of the crack center line, and layer in the depth direction with a dynamic interval on each cross-section; Measure the maximum and minimum widths within each layer and calculate the width volatility within the layer; If the width volatility within the layer is less than the set threshold, calculate the average width change gradient of adjacent layers in sequence along the depth direction.

7. A crack detection method for highway construction according to claim 1, characterized in that: The thermodynamic parameters include the transverse temperature gradient, the longitudinal temperature decay rate, and the heat diffusion area. The specific extraction process includes: Use the Canny edge detection algorithm to identify the crack edge and extract the continuous contour; Select temperature measurement points at equal intervals on both sides of the contour in the long-wave infrared image, and calculate the ratio of the average temperature difference between the two sides to the actual width of the crack to obtain the transverse temperature gradient; Divide the temperature measurement segments at equal intervals along the contour extension direction, and perform linear fitting on the temperature data of each segment to obtain the longitudinal temperature decay rate; Extract the extended area outside the crack contour based on the pre-set temperature boundary, and take the area of the extended area outside the crack contour as the heat diffusion area.

8. A crack detection method for highway construction according to claim 5, characterized in that: The calculation of the comprehensive risk coefficient includes: Calculate the basic risk score according to the crack length and the average crack width, and perform non-linear correction on the score through the crack horizontal development change rate, the crack branch density, and the change gradient of the crack width with depth to obtain the corrected basic risk score; Calculate the thermodynamic score based on the thermodynamic parameters, and fuse it with the corrected basic risk score according to the preset weight to obtain the current comprehensive risk coefficient.

9. A crack detection method for highway construction according to claim 1, characterized in that: The triggering of the graded warning signal includes: Match the comprehensive risk coefficient with the associated comprehensive risk coefficients under each pre-set warning level to obtain the matching warning level as the initial warning level; Obtain the sequence of comprehensive risk coefficients for the current and historical detection periods, and calculate the change rate of the risk coefficient for adjacent periods; Count the number of detection periods in which the risk coefficient continuously increases in the sequence, and calculate its continuous growth ratio to the total number of detection periods; When the change rate of the risk coefficient exceeds the first pre-set warning threshold, or the continuous growth ratio exceeds the second pre-set warning threshold, raise the initial warning level to the next warning level and trigger the warning signal of the next warning level.

10. A crack detection system for highway construction, characterized in that, The system includes the following modules: The multi-spectral thermal imaging acquisition module synchronously acquires multi-band image data on the surface of the asphalt layer through a multi-spectral thermal imaging sensor and performs registration and fusion; The crack interference identification and elimination module eliminates false crack interference data based on the correlation analysis of edge continuity features and thermal radiation fluctuation features and obtains effective multi-band image data; The crack risk analysis module extracts the geometric and thermodynamic parameters of cracks from effective multi-band image data, and calculates the current comprehensive risk coefficient through a weighted scoring model; The crack report generation terminal generates a graded heat map based on the spatial distribution of the current comprehensive risk coefficient, triggers a graded warning signal in combination with the expansion trend, and generates a crack detection report.

Images