Roadway crack identification method and system based on image identification

Abstract

The invention relates to the technical field of image recognition, in particular to a roadway crack recognition method and system based on image recognition. The method comprises the following steps: acquiring roadway structure data; performing regional camera deployment according to the roadway structure data to obtain roadway camera deployment data; performing roadway image acquisition according to the roadway camera deployment data to generate a roadway time sequence image set; performing roadway form change analysis on the roadway time sequence image set to generate roadway time sequence change data; performing surface crack area identification on the roadway time sequence image set through the roadway time sequence change data to generate a roadway surface crack area image; and performing surface crack trend analysis on the roadway surface crack area image to generate surface crack texture trend data. According to the method, through dynamic change detection, hidden crack identification, crack classification and analysis and crack age deduction, the identification accuracy and comprehensiveness are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of image recognition, and in particular to a tunnel crack recognition method and system based on image recognition. Background Art

[0002] Initially, crack detection in tunnels relied on manual inspections. This method is not only time-consuming and labor-intensive, but also easily affected by subjective factors, and cannot guarantee the consistency and accuracy of detection. With the development of computer vision technology, image recognition technology has gradually been introduced into the field of crack detection. Early image recognition methods mainly relied on simple edge detection and image segmentation techniques, which showed certain limitations when dealing with complex crack morphology. After entering the 21st century, breakthroughs in deep learning technology have greatly promoted the development of image recognition technology. The emergence of convolutional neural networks (CNNs) has made automatic feature extraction possible, improving the accuracy and robustness of crack detection. The crack detection system based on CNN can automatically identify various crack types and morphologies by training a large number of annotated images, significantly improving the detection efficiency and accuracy. In addition, with the enhancement of computing power and the application of big data technology, the crack recognition system can process larger data sets and realize real-time detection and monitoring. However, at present, traditional cracks in tunnels usually only focus on surface cracks, ignoring potential hidden cracks, and cannot accurately infer the formation period and age of cracks, which affects the prediction of future development trends of cracks, resulting in low accuracy and comprehensiveness of recognition. Summary of the invention

[0003] Based on this, it is necessary to provide a tunnel crack identification method and system based on image recognition to solve at least one of the above technical problems.

[0004] To achieve the above object, a tunnel crack identification method based on image recognition is provided, the method comprising the following steps: Step S1: Acquire tunnel structure data; perform regional camera deployment according to the tunnel structure data to obtain tunnel camera deployment data; perform tunnel image acquisition according to the tunnel camera deployment data to generate a tunnel time-series image set; perform tunnel morphology change analysis on the tunnel time-series image set to generate tunnel time-series change data; perform surface crack area recognition on the tunnel time-series image set using the tunnel time-series change data to generate a tunnel surface crack area image; Step S2: Analyze the trend of surface cracks on the crack area image of the roadway surface to generate surface crack texture trend data; perform depth map conversion based on the crack area image of the roadway surface to generate a depth map of the crack area of ​​the roadway surface; perform crack ductility analysis on the crack area depth map of the roadway surface to generate surface crack ductility data; confirm the hidden roadway crack area on the crack area image of the roadway surface using the surface crack texture trend data and the surface crack ductility data to obtain a hidden roadway crack area image; Step S3: integrating the tunnel hidden crack region image and the tunnel surface crack region image into a crack type region image to obtain a crack type region image; performing crack thermal balance analysis on the crack type region image to generate tunnel crack thermal balance data; performing crack spectrum analysis on the crack type region image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data; Step S4: Based on the tunnel crack thermal balance data and the tunnel crack spectral analysis data, the crack formation period is divided to generate crack period data; the crack age is deduced and labeled on the crack type area image through the crack period data to generate crack age deduction data; the monitoring frequency of the tunnel camera deployment data is adjusted according to the crack age deduction data to perform dynamic monitoring operations for tunnel crack identification.

[0005] The present invention provides a basis for subsequent camera deployment and image acquisition through accurate tunnel structure data, ensuring the comprehensiveness and accuracy of the data. Cameras are deployed according to tunnel structure data so that image acquisition covers all key areas, improving the coverage of crack detection. By continuously acquiring images to generate a time-series image set, the changing trend of cracks can be tracked to ensure that the dynamic development of cracks is discovered in time. By analyzing the tunnel morphological changes on the time-series image set, the growth, expansion and impact of cracks on the surrounding environment can be identified, improving the accuracy of crack identification. By analyzing the tunnel time-series change data to identify the surface crack area and generate a crack area image, accurate basic data is provided for subsequent depth analysis and hidden crack identification. By analyzing the crack trend data, the direction and mode of crack expansion can be understood, which is helpful to predict the future crack development direction and optimize the maintenance strategy. By converting the crack area image into a depth map, the depth information of the crack can be extracted and potential hidden cracks can be identified, which is often difficult to do in traditional image recognition. The crack ductility analysis of the depth map can understand the expansibility and stability of the crack, providing data support for long-term monitoring and repair of the crack. Combined with the crack texture trend data and crack ductility data, the hidden crack area is confirmed, which improves the comprehensiveness of crack detection and reduces the risk of missed detection. The hidden crack and surface crack area images are integrated to provide a comprehensive crack type image, which is convenient for comprehensive analysis and management. Thermal balance analysis is performed on the crack type area image to reveal the thermal characteristics and formation causes of the cracks and help identify the thermal influencing factors of the cracks. Crack spectrum analysis based on thermal balance data can identify the composition and material properties of the cracks, improve the meticulousness and accuracy of crack detection, and provide guidance for further crack repair and maintenance. The crack formation period is divided using crack thermal balance and spectral analysis data to help understand the historical background and development process of the cracks and support long-term maintenance planning. The age of the cracks is deduced and marked to predict the future development trend of the cracks, optimize maintenance strategies and resource allocation, and reduce potential safety risks. The monitoring frequency of the camera is adjusted according to the crack age deduction data to ensure the dynamic adaptability of the monitoring work, improve the response speed and accuracy to crack changes, and improve the monitoring effect of tunnel cracks. Therefore, the present invention improves the accuracy and comprehensiveness of identification through dynamic change detection, hidden crack identification, crack classification and analysis, and crack age deduction.

[0006] Preferably, step S1 comprises the following steps: Step S11: using laser scanning to obtain tunnel structure data; Step S12: dividing the lane into regions according to the lane structure data to generate lane structure division data; deploying cameras based on the lane structure division data to obtain lane camera deployment data; Step S13: performing lane image acquisition according to the lane camera deployment data at a preset time interval to generate a lane time series image set; performing image preprocessing on the lane time series image set to generate a standard lane time series image set, wherein the image preprocessing includes image filtering, image brightness enhancement and image standardization; Step S14: Analyze the changes in lane morphology of the standard lane time series image set to generate lane time series change data; identify surface crack areas of the standard lane time series image set using the lane time series change data to generate lane surface crack area images.

[0007] The present invention can provide high-precision tunnel structure data through laser scanning technology. These data describe the spatial structure and geometric features of the tunnel in detail, providing a reliable basis for subsequent regional division and analysis. Laser scanning can generate detailed three-dimensional point cloud data, which helps to fully understand the spatial layout of the tunnel and supports accurate regional division and equipment deployment. Regional division is performed according to the tunnel structure data, and the generated tunnel structure division data helps to identify and manage different areas in the tunnel, thereby improving the efficiency of monitoring and management. Camera deployment based on regional division can ensure that important areas and blind areas in the tunnel are fully covered, thereby improving the monitoring effect and the comprehensiveness of data acquisition. The tunnel image set collected at preset time intervals can provide dynamic change information of the tunnel, capture the state at different times, and support subsequent time series analysis and monitoring. Image preprocessing (such as filtering, brightness enhancement and standardization) of the tunnel time series image set can eliminate noise and improve image quality, and generate a standard tunnel time series image set. These standardized image data provide clear and accurate input for subsequent analysis. Morphological change analysis of the standard tunnel time series image set can detect deformation, wear or other structural changes of the tunnel. This helps monitor the health of the tunnel and detect potential problems in a timely manner. The crack area image generated by the tunnel morphology change data can help accurately locate the cracks in the tunnel. This information is crucial for subsequent maintenance and safety assessments. The implementation of the entire step ensures the comprehensiveness and accuracy from data collection to analysis, providing a solid foundation for tunnel monitoring and maintenance. Through the collection and analysis of time-series image sets, the status of the tunnel can be dynamically monitored, and potential problems can be discovered and responded to in a timely manner. The optimization of area division and camera deployment improves the efficiency of tunnel management and the effect of monitoring, and helps to prevent and deal with potential risks in advance.

[0008] Preferably, step S14 comprises the following steps: Step S141: performing feature point recognition on the standard lane time series image set to generate feature points of the standard lane time series image; using the feature points of the standard lane time series image to unify the image perspective of the standard lane time series image set to generate a unified perspective lane image set; Step S142: extracting multi-dimensional morphological features from the unified perspective lane image set to obtain lane multi-dimensional morphological feature data, wherein the multi-dimensional morphological feature extraction includes edge line extraction, surface texture extraction, and geometric contour extraction; calculating the morphological variation of the unified perspective lane image set according to the lane multi-dimensional morphological feature data to generate lane time series morphological variation data; Step S143: using the lane time series morphological change data to segment the lane image set with the uniform view angle into abnormal lane change regions, and generate abnormal lane change region images; performing crack feature matching on the abnormal lane change region images, and generating crack feature matching data; Step S144: annotating the surface crack area of ​​the abnormal tunnel change area image according to the crack feature matching data to generate a tunnel surface crack area image.

[0009] The present invention can accurately extract key feature points in the image by identifying feature points of a standard lane time-series image set. These feature points provide a stable reference, making subsequent image processing more consistent and reliable. The image is unified in perspective using the identified feature points, and the generated unified perspective lane image set eliminates the influence of perspective differences on image analysis. This consistency helps to more accurately compare and analyze lane changes at different time points, and improves the comparability of data. By extracting multidimensional morphological features from the unified perspective lane image set, the edge lines, surface textures and geometric contour information of the lane can be fully obtained. These feature data provide a rich information basis for subsequent morphological change analysis. The morphological change amount is calculated based on the extracted multidimensional morphological feature data, and the degree of change of the lane in different periods can be accurately measured. In this way, small changes in the lane structure can be effectively detected, and potential deformation or damage can be discovered in time. Abnormal change area segmentation using lane time-series morphological change amount data can accurately identify abnormal areas in the lane. This helps to quickly locate areas that need to be focused on and improves the response speed of abnormal situations. Performing crack feature matching on the segmented abnormal area can refine the feature data of the crack and help identify and classify the type and severity of the crack. This can provide accurate data support for subsequent maintenance and repair work. The surface crack area of ​​the abnormal tunnel change area image is marked based on the crack feature matching data. The generated tunnel surface crack area image can clearly identify the location and range of the crack. This is very important for further analysis and decision-making, and helps to formulate targeted maintenance plans.

[0010] Preferably, step S2 comprises the following steps: Step S21: extracting surface crack textures from the crack area image on the tunnel surface to obtain surface crack texture distribution data; performing texture orientation analysis on the surface crack texture distribution data to generate surface crack texture orientation data; performing crack orientation trend analysis on the surface crack texture distribution data according to the surface crack texture orientation data to generate surface crack texture orientation trend data; Step S22: performing depth map conversion based on the crack area image on the tunnel surface to generate a crack area depth map on the tunnel surface; measuring the crack width and crack depth on the crack area depth map on the tunnel surface to obtain surface crack width data and surface crack depth data; Step S23: performing crack ductility analysis based on the surface crack width data and the surface crack depth data to generate surface crack ductility data; Step S24: using the surface crack texture trend data and the surface crack ductility data to confirm the hidden roadway crack area of ​​the roadway surface crack area image, and obtain the hidden roadway crack area image.

[0011] The present invention can obtain detailed texture distribution data of cracks by extracting surface crack textures from crack area images on the surface of the tunnel. This provides basic data for subsequent crack analysis and helps to understand the basic morphology and distribution of cracks. Texture orientation analysis of surface crack texture distribution data can reveal the main direction of crack textures. This information is of great significance for judging the formation mechanism of cracks and their potential impact. Trend analysis of crack orientation is performed on crack orientation according to texture orientation data, and the generated crack texture orientation trend data can help identify the expansion direction and pattern of cracks. This is very useful for predicting the future development direction of cracks and their impact areas. By converting the crack area image on the surface of the tunnel into a depth map, three-dimensional information of the crack can be obtained. This provides accurate spatial data support for the depth and width measurement of cracks. The crack width and depth are measured on the depth map, and the surface crack width data and depth data generated provide detailed geometric parameters of the crack. These data help to assess the severity and safety risks of cracks. Crack ductility analysis is performed through surface crack width data and depth data, and the generated crack ductility data can assess the expansion potential and trend of cracks. This helps predict the scope and speed of crack growth, so that preventive measures can be taken in advance. Using crack texture trend data and crack ductility data to confirm hidden crack areas on the crack area image of the roadway surface can identify crack areas that have not yet appeared or are difficult to detect. This is very important for a comprehensive understanding of the crack situation in the roadway and helps to discover hidden safety hazards. Through crack texture extraction, depth measurement and ductility analysis, the crack situation in the roadway can be fully detected and predicted, providing a scientific basis for subsequent maintenance and management. Detailed crack data, including width, depth and ductility, is provided to help evaluate the impact of cracks on the roadway structure, thereby improving safety. By confirming the hidden crack area, cracks that have not yet appeared can be discovered early, which helps to take timely repair measures and prevent potential safety problems.

[0012] Preferably, step S24 includes the following steps: Step S241: performing crack segment fitting on the crack region image on the surface of the tunnel according to the surface crack ductility data to generate crack extension speed assessment data; Step S242: Perform crack texture line direction vector fitting on the crack area image on the surface of the tunnel according to the surface crack texture trend data to generate crack extension direction assessment data. Step S243: Performing implicit crack association analysis based on the crack extension speed assessment data and the crack extension direction assessment data to generate implicit crack association data; Step S244: Perform stress field simulation on the crack area image on the tunnel surface based on the hidden crack association data to generate a hidden crack area image of the tunnel.

[0013] The present invention can obtain crack extension speed assessment data by performing crack segmentation fitting on the crack area image on the tunnel surface. This helps to understand the expansion speed of the crack, so as to predict the growth of the crack in the future time period. This is crucial to preventing potential structural risks. Crack segmentation fitting provides dynamic monitoring capabilities for crack changes, which helps to timely warn of structural safety hazards, ensure timely maintenance measures, and prevent accidents. Crack extension direction assessment data can be generated by performing crack texture line direction vector fitting based on crack texture trend data. This helps to understand the main expansion direction of the crack and supports the prediction of future development trends of the crack. By identifying the extension direction of the crack, potential hidden crack areas can be accurately located, so that structural maintenance and repair can be performed more effectively. Hidden crack association analysis can be performed through crack extension speed assessment data and crack extension direction assessment data, and hidden cracks that are not visible in the current image but have potential associations with identified cracks can be identified. This is crucial for a comprehensive assessment of the safety of the tunnel structure. Hidden crack association data provides important information for evaluating the overall safety of the structure and helps to formulate more comprehensive maintenance and repair strategies. Based on the hidden crack association data, stress field simulation of the crack area image on the roadway surface can generate hidden crack area images. This provides detailed spatial information for further crack analysis and helps to accurately identify potential hidden crack areas. The stress field simulation results can help formulate specific maintenance strategies for hidden cracks, thereby improving the safety and stability of the roadway structure and preventing the risks caused by crack expansion. By evaluating the crack extension speed and direction, real-time monitoring and prediction of the dynamic changes of cracks can be achieved, and the timeliness of structural safety management can be improved. Through association analysis and stress field simulation, hidden crack areas can be identified, providing detailed data support for structural maintenance and preventing the occurrence of potential safety hazards. It provides a comprehensive assessment of cracks and their expansion trends, which helps to optimize the safety monitoring and maintenance strategies of the roadway and improve the safety and reliability of the overall structure.

[0014] Preferably, step S3 comprises the following steps: Step S31: integrating the tunnel hidden crack region image and the tunnel surface crack region image into a crack type region image to obtain a crack type region image; collecting crack temperature changes in the crack type region image based on infrared thermal imaging technology to generate a tunnel crack temperature change image; Step S32: performing adjacent temperature difference calculation on the tunnel crack temperature change image to obtain the tunnel crack temperature difference data; performing temperature difference change rate calculation on the tunnel crack temperature difference data to obtain the temperature difference change rate data; Step S33: performing crack heat balance analysis on the crack temperature change image of the tunnel using the crack temperature difference data and the temperature difference change rate data to generate crack heat balance data of the tunnel; Step S34: Based on the spectral analysis technology, crack spectrum analysis is performed on the crack type area image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data.

[0015] The present invention can obtain a detailed crack type area image by integrating the hidden crack area image of the tunnel with the crack area image of the tunnel surface. This integration helps to fully understand the distribution of various types of cracks in the tunnel. By using infrared thermal imaging technology to collect temperature changes in the crack type area image, a tunnel crack temperature change image can be generated. This helps to detect temperature changes in the crack part, provide information on the thermal state of the crack, and assist in evaluating the thermal anomaly of the crack. By calculating the adjacent temperature difference of the tunnel crack, the tunnel crack temperature difference data is obtained. The temperature difference data helps to identify whether there is a significant thermal anomaly at the crack, which indicates the potential problem or change trend of the crack. The temperature difference change rate calculation can generate temperature difference change rate data. By evaluating the temperature difference change rate, the dynamic changes of the crack thermal balance can be understood, the aggravation or relief of the temperature difference can be identified, and it is helpful to better understand the thermal behavior of the crack. The crack thermal balance analysis can be performed by the tunnel crack temperature difference data and the temperature difference change rate data, and the tunnel crack thermal balance data can be generated. This helps to evaluate the stability of the crack in the thermal environment and identify whether there is a thermal imbalance problem. Fracture heat balance data can be used to identify abnormal thermal phenomena and prevent further expansion or structural damage of cracks due to changes in thermal stress. Fracture spectral analysis of crack type area images based on spectral analysis technology can generate tunnel crack spectral analysis data. This helps to identify the chemical composition and physical characteristics of cracks and provide in-depth information about the properties of cracks. Fracture spectral analysis data helps to diagnose the type of cracks, the causes of their formation and their impact on the overall structure of the tunnel, and supports the formulation of targeted repair and maintenance strategies. By combining different technologies (such as infrared thermal imaging and spectral analysis), the thermal state and chemical properties of cracks can be comprehensively evaluated to improve the accuracy of crack monitoring. Temperature difference calculation and change rate analysis provide dynamic information on the thermal behavior of cracks, which helps to monitor the changing trends of cracks and detect potential problems in a timely manner. Fracture heat balance analysis and spectral analysis provide a scientific basis for the formulation of effective maintenance and repair strategies, and improve the safety and stability of tunnel structures.

[0016] Preferably, step S34 includes the following steps: Step S341: performing Raman spectrum acquisition on the crack type area image according to the roadway crack thermal balance data based on the spectrum analysis technology to obtain the roadway crack spectrum map; Step S342: performing crack environment element analysis on the crack spectrum of the roadway to generate crack environment element data; identifying chemical composition of the crack type area data through the crack environment element data to generate crack environment chemical substance data; Step S343: analyzing the oxidation state of the crack material on the crack spectrum of the roadway to generate crack material oxidation state data; analyzing the oxidation degree of the crack type area data according to the crack material oxidation state data to generate crack oxidation degree data; Step S344: Analyze the changes in crack spectrum characteristics during the time period of the roadway crack spectrum diagram using the crack environmental chemical substance data and the crack oxidation degree data to generate roadway crack spectrum analysis data.

[0017] The tunnel crack spectrum map generated by the present invention through Raman spectroscopy acquisition provides detailed spectral information of the crack area. This helps to accurately identify the material composition and chemical structure in the crack and reveal the microscopic characteristics of the crack. Raman spectroscopy technology can detect the vibration mode of the crack and provide high-resolution spectral data for the chemical and physical properties of the crack material. This supports a deep understanding of the causes of the formation of the crack and the material characteristics. Through the crack environmental element analysis, the crack environmental element data generated helps to identify the environmental components of the crack area, such as minerals and pollutants. This is very critical to understanding the formation and expansion mechanism of the crack. The environmental element data is used to identify the chemical composition and generate the crack environmental chemical data. It can reveal the chemical composition in the crack, evaluate the impact of the environment on the crack, and support accurate maintenance and control measures. The crack material oxidation state data generated by the crack material oxidation state analysis provides information on the degree of oxidation of the crack material. This is crucial to evaluate the degree of degradation and corrosion of the crack material. Through the oxidation degree analysis, the impact of crack oxidation on the structure can be evaluated, which helps predict the further development of the crack and the impact on the overall structure, and formulate appropriate repair strategies. Through the crack environmental chemical data and crack oxidation degree data, the crack spectral characteristics are analyzed for time period changes to generate tunnel crack spectral analysis data. It is possible to monitor the changing trend of crack characteristics over time and provide a dynamic understanding of crack progression. Analyzing the time period changes in crack spectral characteristics helps to comprehensively evaluate the evolution of cracks and identify potential risk factors, thereby supporting more scientific maintenance and repair decisions. Combining Raman spectroscopy data and environmental analysis, the chemical and physical properties of cracks can be accurately diagnosed, providing a scientific basis for crack assessment and repair. The time period change analysis of spectral characteristics makes it possible to assess the dynamic risks of cracks and helps to foresee the future development trend of cracks. By comprehensively analyzing the environmental impact and material status of cracks, it supports the formulation of effective maintenance and repair plans and improves the safety and stability of tunnel structures.

[0018] Preferably, step S4 comprises the following steps: Step S41: dividing the crack formation period based on the roadway crack heat balance data and the roadway crack spectrum analysis data to generate crack period data; Step S42: deduce and mark the fracture age of the fracture type area image using the fracture period data to generate fracture age deduction data; Step S43: Compare the fracture age deduction data with the preset fracture age range. When the fracture age deduction data is less than the preset fracture age range, the corresponding fracture type regional image is marked as a fracture new formation period image; when the fracture age deduction data is within the preset fracture age range, the corresponding fracture type regional image is marked as a fracture stable period image; when the fracture age deduction data is greater than the preset fracture age range, the corresponding fracture type regional image is marked as a fracture aging period image; Step S44: Perform crack age image discrimination on the crack type area annotated image. When it is confirmed that the crack type area image is a crack newly formed period image, high-frequency monitoring of the alley camera deployment data is performed. When it is confirmed that the crack type area image is a crack stable period image, medium-frequency monitoring of the alley camera deployment data is performed. When it is confirmed that the crack type area image is a crack aging period image, low-frequency monitoring of the alley camera deployment data is performed to execute dynamic monitoring operations for alley crack identification.

[0019] The present invention can accurately determine the formation time of the cracks by dividing the crack formation period based on the crack thermal balance data and spectral analysis data. This is crucial to understanding the evolution of the cracks and their impact on the structure. The generated crack period data provides a classification basis for the time dimension for crack management, which is helpful to formulate maintenance and treatment strategies at different stages. The generated crack period data provides a classification basis for the time dimension for crack management, which is helpful to formulate maintenance and treatment strategies at different stages. The generated crack period data provides a classification basis for the time dimension for crack management, which is helpful to formulate maintenance and treatment strategies at different stages. By comparing the crack age deduction data with the preset age range, the cracks can be classified into the new formation period, stable period or aging period. This classification helps to assess the risk level of the cracks and their impact on structural safety. Classification based on the age stage of the cracks can give priority to high-risk cracks in maintenance and repair work to ensure the most effective use of resources. According to the age stage of the crack type area image, the monitoring frequency of the lane camera is adjusted. High-frequency monitoring of newly formed cracks can track their development and changes in real time; medium-frequency monitoring of stable cracks can keep a constant eye on their status; low-frequency monitoring of aging cracks can reduce unnecessary monitoring work. This dynamic adjustment helps optimize the allocation of monitoring resources. By implementing different monitoring frequencies for different crack age stages, monitoring efficiency can be improved, redundant data collection can be reduced, and potential risks can be warned more effectively. According to the age and development stage of the cracks, targeted maintenance and repair measures can be formulated to improve the safety and stability of the roadway. Through classification and staged monitoring, the risks of cracks can be effectively identified and controlled to ensure the structural safety of the roadway. Dynamically adjust the monitoring frequency, optimize resource allocation according to the actual status of the cracks, and improve the efficiency and effectiveness of monitoring. Through classification and staged monitoring, the risks of cracks can be effectively identified and controlled to ensure the structural safety of the roadway.

[0020] Preferably, step S42 includes the following steps: Step S421: extracting crack period features from the crack period data using a preset crack period feature database to obtain crack period feature data; Step S422: dividing the data set of the crack period characteristic data to generate a model training set and a model test set; using the support vector machine algorithm to perform model training on the model training set to generate a crack age deduction pre-model; Step S423: The fracture age deduction pre-model is optimized and iterated through the model test set to generate a fracture age deduction model; the fracture type area image is imported into the fracture age deduction model to perform fracture age deduction and annotation to generate a fracture type area annotation image.

[0021] The present invention extracts crack period feature data through a preset crack period feature database, which helps to accurately extract the time characteristics of the crack, thereby providing more detailed information for subsequent analysis. Feature extraction provides a deep understanding of the crack period, lays a solid foundation for model training and deduction, and improves the prediction accuracy of the model. The crack period feature data is divided into a model training set and a model test set, and trained using a support vector machine algorithm to generate an efficient crack age deduction pre-model. The support vector machine algorithm has excellent classification ability and generalization performance, and can handle complex data patterns. The crack age deduction pre-model is optimized and iterated through the model test set to ensure that the model can accurately reflect the actual situation, and effectively deduce different crack ages, thereby improving the reliability and prediction accuracy of the model. The crack type area image is imported into the crack age deduction model, which can realize the accurate deduction and annotation of the crack age. This method not only improves the accuracy of deduction, but also provides a scientific basis for subsequent maintenance and repair work. The generated crack type area annotation image provides a clear crack age classification, which helps to quickly identify cracks of different age groups, and is convenient for hierarchical management and processing of cracks. Through scientific feature extraction and model training, the accuracy of crack age deduction can be improved, providing reliable data support for crack management in tunnels. Accurate crack age deduction helps to formulate more effective maintenance strategies, prioritize according to the actual age of cracks, and improve the pertinence and efficiency of maintenance work. Providing detailed crack age data enables decision makers to make scientific and reasonable decisions based on data, thereby optimizing the safety management and maintenance of tunnels.

[0022] In this specification, a tunnel crack identification system based on image recognition is provided, which is used to execute the above-mentioned tunnel crack identification method based on image recognition. The tunnel crack identification system based on image recognition includes: The surface crack area recognition module is used to obtain the tunnel structure data; perform regional camera deployment according to the tunnel structure data to obtain the tunnel camera deployment data; perform tunnel image acquisition according to the tunnel camera deployment data to generate a tunnel time series image set; perform tunnel morphology change analysis on the tunnel time series image set to generate tunnel time series change data; perform surface crack area recognition on the tunnel time series image set through the tunnel time series change data to generate a tunnel surface crack area image; The hidden crack area identification module is used to perform surface crack trend analysis on the crack area image on the roadway surface to generate surface crack texture trend data; perform depth map conversion based on the crack area image on the roadway surface to generate a crack area depth map on the roadway surface; perform crack ductility analysis on the crack area depth map on the roadway surface to generate surface crack ductility data; use the surface crack texture trend data and surface crack ductility data to confirm the hidden roadway crack area on the crack area image on the roadway surface to obtain a hidden roadway crack area image; The crack feature recognition module is used to integrate the tunnel hidden crack area image and the tunnel surface crack area image into a crack type area image to obtain a crack type area image; perform crack thermal balance analysis on the crack type area image to generate tunnel crack thermal balance data; perform crack spectrum analysis on the crack type area image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data; The monitoring frequency adjustment module is used to divide the crack formation period based on the tunnel crack thermal balance data and the tunnel crack spectral analysis data to generate crack period data; to deduce and mark the crack age of the crack type area image through the crack period data to generate crack age deduction data; and to adjust the monitoring frequency of the tunnel camera deployment data according to the crack age deduction data to perform dynamic monitoring operations for tunnel crack identification.

[0023] The beneficial effect of the present invention is that by acquiring the tunnel structure data and deploying cameras, the structural layout of the tunnel can be fully understood, ensuring that data collection covers the entire tunnel area, and ensuring the comprehensiveness and accuracy of subsequent analysis. Analyzing the tunnel time-series image set can monitor the changes in the tunnel morphology, identify potential surface crack areas, and provide key data support for the timely discovery and treatment of cracks. Through morphological change analysis and surface crack area identification, the crack area can be effectively distinguished, laying the foundation for subsequent crack feature analysis and evaluation. Analyzing the trend of surface cracks can understand the expansion direction of the cracks and provide valuable information for further crack management and repair. Through depth map conversion and crack ductility analysis, the width and depth of the cracks can be accurately measured, the expansibility of the cracks can be evaluated, and the degree of impact of the cracks on the tunnel structure can be evaluated. Combining crack trend data and ductility data to confirm the hidden crack area helps to discover those cracks that are not easy to detect, so as to carry out maintenance in advance and avoid potential risks. Integrating the hidden crack area image with the surface crack area image provides a comprehensive crack type area image, which helps to better understand the distribution and properties of the cracks. Through the crack heat balance analysis, the thermal characteristics of the cracks can be understood, which helps to reveal the formation mechanism of the cracks and their impact on the tunnel structure. Spectral analysis of the cracks can detect the chemical composition and material properties in the cracks, provide an in-depth understanding of the causes of the cracks, and provide a scientific basis for formulating repair strategies. According to the crack heat balance data and spectral analysis data, the crack formation period can be divided, the age of the cracks can be determined, the evolution process of the cracks can be understood, and time-effective information can be provided for crack management. Through crack age deduction and labeling, the age status of different cracks can be accurately understood, so as to formulate corresponding maintenance strategies according to the actual situation of the cracks. According to the crack age deduction data, the monitoring frequency of the tunnel camera can be adjusted to achieve targeted dynamic monitoring, improve monitoring efficiency and effect, and ensure timely discovery and handling of new crack problems. Therefore, the present invention improves the accuracy and comprehensiveness of identification through dynamic change detection, hidden crack identification, crack classification and analysis, and crack age deduction. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 A schematic diagram of the steps of a method for identifying cracks in a tunnel based on image recognition; Figure 2 for Figure 1 Detailed implementation steps of step S2 in the flowchart; Figure 3 for Figure 1 Detailed implementation steps of step S3 in FIG. Figure 4 for Figure 1 Detailed implementation steps of step S4 in FIG. The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0025] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0026] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0027] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0028] To achieve this, please refer to Figures 1 to 4 , a tunnel crack identification method based on image recognition, the method comprising the following steps: Step S1: Acquire tunnel structure data; perform regional camera deployment according to the tunnel structure data to obtain tunnel camera deployment data; perform tunnel image acquisition according to the tunnel camera deployment data to generate a tunnel time-series image set; perform tunnel morphology change analysis on the tunnel time-series image set to generate tunnel time-series change data; perform surface crack area recognition on the tunnel time-series image set using the tunnel time-series change data to generate a tunnel surface crack area image; Step S2: Analyze the trend of surface cracks on the crack area image of the roadway surface to generate surface crack texture trend data; perform depth map conversion based on the crack area image of the roadway surface to generate a depth map of the crack area of ​​the roadway surface; perform crack ductility analysis on the crack area depth map of the roadway surface to generate surface crack ductility data; confirm the hidden roadway crack area on the crack area image of the roadway surface using the surface crack texture trend data and the surface crack ductility data to obtain a hidden roadway crack area image; Step S3: integrating the tunnel hidden crack region image and the tunnel surface crack region image into a crack type region image to obtain a crack type region image; performing crack thermal balance analysis on the crack type region image to generate tunnel crack thermal balance data; performing crack spectrum analysis on the crack type region image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data; Step S4: Based on the tunnel crack thermal balance data and the tunnel crack spectral analysis data, the crack formation period is divided to generate crack period data; the crack age is deduced and labeled on the crack type area image through the crack period data to generate crack age deduction data; the monitoring frequency of the tunnel camera deployment data is adjusted according to the crack age deduction data to perform dynamic monitoring operations for tunnel crack identification.

[0029] The present invention provides a basis for subsequent camera deployment and image acquisition through accurate tunnel structure data, ensuring the comprehensiveness and accuracy of the data. Cameras are deployed according to tunnel structure data so that image acquisition covers all key areas, improving the coverage of crack detection. By continuously acquiring images to generate a time-series image set, the changing trend of cracks can be tracked to ensure that the dynamic development of cracks is discovered in time. By analyzing the tunnel morphological changes on the time-series image set, the growth, expansion and impact of cracks on the surrounding environment can be identified, improving the accuracy of crack identification. By analyzing the tunnel time-series change data to identify the surface crack area and generate a crack area image, accurate basic data is provided for subsequent depth analysis and hidden crack identification. By analyzing the crack trend data, the direction and mode of crack expansion can be understood, which is helpful to predict the future crack development direction and optimize the maintenance strategy. By converting the crack area image into a depth map, the depth information of the crack can be extracted and potential hidden cracks can be identified, which is often difficult to do in traditional image recognition. The crack ductility analysis of the depth map can understand the expansibility and stability of the crack, providing data support for long-term monitoring and repair of the crack. Combined with the crack texture trend data and crack ductility data, the hidden crack area is confirmed, which improves the comprehensiveness of crack detection and reduces the risk of missed detection. The hidden crack and surface crack area images are integrated to provide a comprehensive crack type image, which is convenient for comprehensive analysis and management. Thermal balance analysis is performed on the crack type area image to reveal the thermal characteristics and formation causes of the cracks and help identify the thermal influencing factors of the cracks. Crack spectrum analysis based on thermal balance data can identify the composition and material properties of the cracks, improve the meticulousness and accuracy of crack detection, and provide guidance for further crack repair and maintenance. The crack formation period is divided using crack thermal balance and spectral analysis data to help understand the historical background and development process of the cracks and support long-term maintenance planning. The age of the cracks is deduced and marked to predict the future development trend of the cracks, optimize maintenance strategies and resource allocation, and reduce potential safety risks. The monitoring frequency of the camera is adjusted according to the crack age deduction data to ensure the dynamic adaptability of the monitoring work, improve the response speed and accuracy to crack changes, and improve the monitoring effect of tunnel cracks. Therefore, the present invention improves the accuracy and comprehensiveness of identification through dynamic change detection, hidden crack identification, crack classification and analysis, and crack age deduction.

[0030] In the embodiment of the present invention, reference Figure 1 The above is a schematic diagram of the steps of a method for identifying cracks in a tunnel based on image recognition according to the present invention. In this example, the method for identifying cracks in a tunnel based on image recognition includes the following steps: Step S1: Acquire tunnel structure data; perform regional camera deployment according to the tunnel structure data to obtain tunnel camera deployment data; perform tunnel image acquisition according to the tunnel camera deployment data to generate a tunnel time-series image set; perform tunnel morphology change analysis on the tunnel time-series image set to generate tunnel time-series change data; perform surface crack area recognition on the tunnel time-series image set using the tunnel time-series change data to generate a tunnel surface crack area image; In an embodiment of the present invention, the geometric structure, geological features and other relevant information of the tunnel are obtained by laser scanning, geological survey or existing engineering drawings. The collected raw data is converted into a structured format, such as a 3D model or a digital topographic map, to ensure the accuracy and completeness of the data. According to the structural characteristics of the tunnel and the key monitoring areas, the optimal position of the camera is planned to ensure that all areas that need to be monitored are covered. The camera is installed at the planned location, including fixtures and cable wiring, to ensure that the camera can work stably and has a good viewing angle. The shooting parameters of the camera, including resolution, frame rate, exposure, etc., are set to adapt to the lighting and environmental conditions of the tunnel. The camera is started for continuous or timed shooting to generate a tunnel time series image set. As needed, image acquisition can be performed at different time points to capture changes in the tunnel. The tunnel time series image set is preprocessed, such as denoising, contrast enhancement, distortion correction, etc. Image processing algorithms (such as background modeling, change detection algorithms) are applied to analyze changes in the image sequence to generate tunnel time series change data. Common methods include differential images, optical flow analysis, etc. Use deep learning or traditional image processing techniques (such as edge detection and region growing) to analyze the tunnel time series image set and identify the surface crack areas. Convolutional neural networks (CNNs) can be used for more accurate crack detection, generate images of crack areas on the tunnel surface, highlight the crack areas, and provide data support for subsequent analysis and processing.

[0031] Step S2: Analyze the trend of surface cracks on the crack area image of the roadway surface to generate surface crack texture trend data; perform depth map conversion based on the crack area image of the roadway surface to generate a depth map of the crack area of ​​the roadway surface; perform crack ductility analysis on the crack area depth map of the roadway surface to generate surface crack ductility data; confirm the hidden roadway crack area on the crack area image of the roadway surface using the surface crack texture trend data and the surface crack ductility data to obtain a hidden roadway crack area image; In an embodiment of the present invention, the crack region image on the tunnel surface is preprocessed, including image denoising, contrast enhancement, etc., to improve the detection accuracy of the crack direction. The crack texture features are extracted using texture analysis algorithms (such as grayscale co-occurrence matrix, local binary pattern, etc.). The crack direction trend is analyzed by using methods such as directional gradient histogram (HOG) and edge detection to generate surface crack texture trend data. The crack direction detection results can be further optimized in combination with machine learning methods (such as SVM and decision tree). Based on the crack region image on the tunnel surface, a depth map of the crack region on the tunnel surface is generated using stereo vision technology or deep learning methods (such as monocular depth estimation network). Stereo vision technology requires at least two images at different angles for depth calculation, while deep learning methods estimate depth directly from a single image through a trained network. The generated depth map is post-processed, such as smoothing, noise removal, and interpolation, to improve the quality of the depth map. The geometric features of the cracks are extracted from the depth map, including the width, depth, length, etc. of the cracks. Apply geometric analysis algorithms (such as crack propagation models and fracture mechanics models) to analyze the ductility of cracks and generate surface crack ductility data, including the potential expansion direction and distribution range of cracks. Combine the surface crack texture trend data and surface crack ductility data, and use data fusion technology (such as weighted average and decision fusion) to conduct in-depth analysis of the crack area on the tunnel surface. By analyzing the crack direction and ductility, the existing hidden crack areas are identified. These areas are not easy to find in existing images, but their existence can be confirmed through data analysis, and hidden tunnel crack area images are generated to highlight potential hidden crack areas and provide detailed crack propagation information.

[0032] Step S3: integrating the tunnel hidden crack region image and the tunnel surface crack region image into a crack type region image to obtain a crack type region image; performing crack thermal balance analysis on the crack type region image to generate tunnel crack thermal balance data; performing crack spectrum analysis on the crack type region image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data; In an embodiment of the present invention, by preparing the hidden crack area image of the tunnel and the crack area image of the surface of the tunnel, their resolution and alignment are ensured to be consistent. The image can be aligned using image registration technology. The crack area is identified using an image processing algorithm (such as region segmentation, edge detection), and the hidden crack area and the surface crack area are classified and marked. The identified crack areas are merged into a comprehensive crack type area image. Different types of crack area images can be merged using image fusion technology (such as weighted fusion, pixel-level fusion) to generate a complete crack type area image. The tunnel is photographed using a thermal imaging device to obtain thermal image data of the crack area. These data can reflect the thermal characteristics and thermal distribution of the crack. A thermal analysis model (such as a heat conduction model, a heat balance equation) is applied to perform thermal balance analysis on the crack type area image. The thermal distribution, heat flux density and thermal balance state of the crack area are calculated to generate the thermal balance data of the crack in the tunnel. The consistency of the results of the thermal balance analysis with the actual observation data is verified to ensure the accuracy of the data. A spectrometer or a hyperspectral imaging device is used to collect spectra of the crack area in the tunnel to obtain spectral data of the crack area. These data include information such as reflectivity and absorptivity at different wavelengths. The collected spectral data are analyzed, and spectral characteristics of the cracks are extracted using spectral analysis methods (such as principal component analysis and spectral feature extraction). This can help identify the composition and mineral content of the crack material. The spectral data are aligned with the crack type area image to generate tunnel crack spectral analysis data. These data can reveal the composition characteristics and possible chemical changes of the cracks.

[0033] Step S4: Based on the tunnel crack thermal balance data and the tunnel crack spectral analysis data, the crack formation period is divided to generate crack period data; the crack age is deduced and labeled on the crack type area image through the crack period data to generate crack age deduction data; the monitoring frequency of the tunnel camera deployment data is adjusted according to the crack age deduction data to perform dynamic monitoring operations for tunnel crack identification.

[0034] In an embodiment of the present invention, the thermal balance data of the roadway cracks and the spectral analysis data of the roadway cracks are fused, and these data are used to infer the formation period of the cracks. The thermal characteristics and spectral characteristics of the cracks can reveal the time information of the formation and evolution of the cracks. The time series analysis method and the thermal spectrum model are applied to identify the formation period of the cracks. According to the change pattern in the thermal balance data and the spectral data, the age or formation period of the cracks is estimated. The cracks are divided into different stages or time periods according to their formation period, and the crack period data is generated. These periods can be defined based on the changes in the thermal stability of the cracks, the changes in the spectral characteristics, etc. A crack age deduction model is constructed, and the crack period data is applied to the model. This can use statistical methods or machine learning models (such as regression analysis, classification models) to predict the age of the cracks. The crack type area image is annotated according to the crack age deduction data. The age information of the cracks is annotated into the image using an image annotation tool to generate crack age deduction data. The crack age deduction results are verified to ensure their accuracy and reliability. It can be verified by comparing with the results of on-site inspections or using the age of known cracks as a basis. Based on the fracture age deduction data, analyze the monitoring needs of fractures of different ages. For example, newer fractures may require more frequent monitoring, while older fractures may require less frequent monitoring. Adjust the monitoring frequency of tunnel cameras based on the fracture age deduction data. Develop a monitoring frequency adjustment strategy, such as increasing the monitoring frequency of new fracture areas and reducing the monitoring frequency of old fracture areas. Implement dynamic monitoring operations, configure cameras according to the adjusted monitoring frequency, and track the changes and development of fractures in real time.

[0035] Preferably, step S1 comprises the following steps: Step S11: using laser scanning to obtain tunnel structure data; Step S12: dividing the lane into regions according to the lane structure data to generate lane structure division data; deploying cameras based on the lane structure division data to obtain lane camera deployment data; Step S13: performing lane image acquisition according to the lane camera deployment data at a preset time interval to generate a lane time series image set; performing image preprocessing on the lane time series image set to generate a standard lane time series image set, wherein the image preprocessing includes image filtering, image brightness enhancement and image standardization; Step S14: Analyze the changes in lane morphology of the standard lane time series image set to generate lane time series change data; identify surface crack areas of the standard lane time series image set using the lane time series change data to generate lane surface crack area images.

[0036] In an embodiment of the present invention, a laser scanning device is used to perform high-precision measurement of the interior of the lane to obtain detailed lane structure data. These data include geometric features such as the width, height, shape, and wall inclination of the lane, as well as obstacles and surface details in the lane. Based on the acquired lane structure data, the lane is reasonably divided into regions. The division basis specifically includes the geometric shape of the lane, key positions (such as bends, intersections), or functional areas are divided according to different monitoring requirements. According to the divided areas, combined with the coverage range of the camera and the monitoring target, the optimal deployment position of the camera in each area is determined to generate lane camera deployment data. These data ensure that the camera can fully cover the key areas in the lane and minimize the monitoring blind spots. Image acquisition is performed according to the set time interval (such as every hour, every day), the state of the lane at different time points is recorded, and a lane time series image set is generated. Noise in the time series image is removed to maintain the clarity of the image. The image brightness is adjusted to ensure that the details of the lane can be clearly observed under different lighting conditions. The images are standardized to eliminate the differences caused by different shooting conditions (such as lighting and angles), so that the time-series images are comparable and a standard tunnel time-series image set is generated. Image analysis technology is used to analyze the changes in tunnel morphology in the standard tunnel time-series image set. For example, by comparing images at different times, the shape changes, ground subsidence, wall displacement and other phenomena inside the tunnel are detected. The analysis results are recorded in the form of tunnel time-series change data. Based on the tunnel time-series change data, the crack areas that may appear on the tunnel surface are further identified. This process includes steps such as crack detection, boundary recognition and position marking, and finally generates images of crack areas on the tunnel surface. These images can be used for subsequent maintenance and repair work.

[0037] Preferably, step S14 comprises the following steps: Step S141: performing feature point recognition on the standard lane time series image set to generate feature points of the standard lane time series image; using the feature points of the standard lane time series image to unify the image perspective of the standard lane time series image set to generate a unified perspective lane image set; Step S142: extracting multi-dimensional morphological features from the unified perspective lane image set to obtain lane multi-dimensional morphological feature data, wherein the multi-dimensional morphological feature extraction includes edge line extraction, surface texture extraction, and geometric contour extraction; calculating the morphological variation of the unified perspective lane image set according to the lane multi-dimensional morphological feature data to generate lane time series morphological variation data; Step S143: using the lane time series morphological change data to segment the lane image set with the uniform view angle into abnormal lane change regions, and generate abnormal lane change region images; performing crack feature matching on the abnormal lane change region images, and generating crack feature matching data; Step S144: annotating the surface crack area of ​​the abnormal tunnel change area image according to the crack feature matching data to generate a tunnel surface crack area image.

[0038] In the embodiment of the present invention, the images in the standard lane time series image set are processed by using a feature point recognition algorithm to identify key feature points in the image, such as corners, intersections and obvious texture features of the lane wall. By matching the image feature points at different times, the shooting angle of the image is corrected, the influence caused by the change of camera position or angle is eliminated, and all images are compared and analyzed at the same angle to generate a unified angle lane image set. The edge lines in the lane, such as the boundary between the wall, the ground and the ceiling, are identified and extracted to obtain the contour information of the lane. The texture features of the lane surface are extracted to identify the texture and subtle changes of the material. The geometric contour of the lane, including the shape and structural information of the wall, the ground and the ceiling, is extracted to generate the overall contour feature data of the lane. By comparing the morphological features at different times, the morphological change amount of the lane in the time series, such as the displacement of the wall, the degree of damage on the surface, etc., is calculated to generate the lane time series morphological change amount data. Based on the lane time series morphological change amount data, the areas in the lane where abnormal changes occur are identified. These areas may have drastic changes in morphology, indicating potential structural problems or damage. Further crack feature matching is performed on the identified abnormal areas. By comparing the morphological features of the abnormal areas, possible cracks are identified and crack feature matching data is generated. The crack feature matching data is used to accurately mark the cracks in the abnormal change area and generate detailed crack area images on the roadway surface. These images can clearly show the location, size and morphology of the cracks, providing a basis for subsequent repair and reinforcement work.

[0039] As an example of the present invention, refer to Figure 2 As shown, in this example, step S2 includes: Step S21: extracting surface crack textures from the crack area image on the tunnel surface to obtain surface crack texture distribution data; performing texture orientation analysis on the surface crack texture distribution data to generate surface crack texture orientation data; performing crack orientation trend analysis on the surface crack texture distribution data according to the surface crack texture orientation data to generate surface crack texture orientation trend data; Step S22: performing depth map conversion based on the crack area image on the tunnel surface to generate a crack area depth map on the tunnel surface; measuring the crack width and crack depth on the crack area depth map on the tunnel surface to obtain surface crack width data and surface crack depth data; Step S23: performing crack ductility analysis based on the surface crack width data and the surface crack depth data to generate surface crack ductility data; Step S24: using the surface crack texture trend data and the surface crack ductility data to confirm the hidden roadway crack area of ​​the roadway surface crack area image, and obtain the hidden roadway crack area image.

[0040] In the embodiment of the present invention, the cracks in the crack area image of the roadway surface are extracted by using an image processing algorithm, and the texture features of the cracks, such as the direction, shape, spacing, etc. of the cracks, are identified to obtain the surface crack texture distribution data. The extracted crack texture is analyzed for orientation, the overall direction and inclination angle of the crack are calculated, and the surface crack texture orientation data is generated. Combined with the crack texture orientation data, the crack trend is analyzed, the direction in which the crack may expand is identified, and the surface crack texture orientation trend data is generated. Using the depth information of the image, the crack area image of the roadway surface is converted into a depth image, the three-dimensional structure of the crack is displayed, and the depth map of the crack area on the roadway surface is generated. The width and depth of the crack are measured on the depth map, the size and depth of the crack are accurately evaluated, and the surface crack width data and surface crack depth data are generated. Based on the width and depth data of the crack, the ductility of the crack is analyzed, the possibility and rate of expansion of the crack in the roadway are evaluated, and the surface crack ductility data is generated. This analysis can reveal the possible development trend of the cracks in different locations and predict their possible changes in the future. Combined with the texture trend data and ductility data of surface cracks, the potential expansion path and stress condition of the cracks are further analyzed to identify the hidden crack areas that may exist in the tunnel but have not yet appeared. These hidden crack areas are marked in the image to generate hidden tunnel crack area images for subsequent monitoring and maintenance.

[0041] Preferably, step S24 includes the following steps: Step S241: performing crack segment fitting on the crack region image on the surface of the tunnel according to the surface crack ductility data to generate crack extension speed assessment data; Step S242: Perform crack texture line direction vector fitting on the crack area image on the surface of the tunnel according to the surface crack texture trend data to generate crack extension direction assessment data. Step S243: Performing implicit crack association analysis based on the crack extension speed assessment data and the crack extension direction assessment data to generate implicit crack association data; Step S244: Perform stress field simulation on the crack area image on the tunnel surface based on the hidden crack association data to generate a hidden crack area image of the tunnel.

[0042] In an embodiment of the present invention, the cracks in the crack area image of the tunnel surface are segmented by using the surface crack ductility data. First, the crack area is divided into several sections, and polynomial fitting or Bezier curve fitting is performed according to the morphological characteristics of each crack section to obtain the ductility data of each crack section. The fitting process can accurately capture the subtle changes of the cracks. By comparing the crack fitting results of different time periods, the extension speed of the cracks in each section is calculated. The extension speed evaluation can be combined with time series analysis technology to generate crack extension speed evaluation data for subsequent crack development trend prediction. Using the surface crack texture trend data, the direction of the crack texture line in the crack area image of the tunnel surface is fitted. By analyzing the microstructure and direction changes of the crack texture, the vector analysis method is used to fit the main direction vector of the crack texture line. According to the fitted direction vector, combined with the crack extension speed evaluation data, the main extension direction of the crack is determined. This direction evaluation can be quantified in combination with the spatial coordinate system to generate crack extension direction evaluation data. The crack extension speed assessment data and the crack extension direction assessment data are comprehensively analyzed to identify potential hidden cracks. Hidden cracks refer to cracks that have not yet been fully revealed due to local stress concentration or material defects. Through correlation analysis, the potential connection or expansion trend between cracks is found, and hidden crack correlation data is generated. This data can be used to predict the future damage risk of tunnel structures. Based on the hidden crack correlation data, the stress field simulation of the crack area on the tunnel surface is carried out. The stress field simulation can be performed by finite element analysis (FEA) or other numerical simulation methods to simulate the stress distribution of the tunnel under different load conditions. The simulation focuses on the stress concentration degree of the hidden crack area to determine its possible expansion path. The potential hidden crack area is marked by the stress field simulation results. Combined with the previous crack extension speed and direction assessment data, the final hidden crack area image of the tunnel is generated. This image can be used to formulate a tunnel maintenance plan to avoid structural failure caused by further expansion of hidden cracks.

[0043] As an example of the present invention, refer to Figure 3 As shown, in this example, step S3 includes: Step S31: integrating the tunnel hidden crack region image and the tunnel surface crack region image into a crack type region image to obtain a crack type region image; collecting crack temperature changes in the crack type region image based on infrared thermal imaging technology to generate a tunnel crack temperature change image; Step S32: performing adjacent temperature difference calculation on the tunnel crack temperature change image to obtain the tunnel crack temperature difference data; performing temperature difference change rate calculation on the tunnel crack temperature difference data to obtain the temperature difference change rate data; Step S33: performing crack heat balance analysis on the crack temperature change image of the tunnel using the crack temperature difference data and the temperature difference change rate data to generate crack heat balance data of the tunnel; Step S34: Based on the spectral analysis technology, crack spectrum analysis is performed on the crack type area image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data.

[0044] In an embodiment of the present invention, the image of the hidden crack area of ​​the tunnel and the image of the surface crack area of ​​the tunnel are integrated. The integration process can be achieved through image superposition, feature matching and regional fusion technology to ensure that the hidden cracks and surface cracks are fully displayed in the same image. The crack type area image finally generated will contain all known crack types and positions in the tunnel. Based on infrared thermal imaging technology, the temperature change of the integrated crack type area image is monitored. The infrared thermal imaging device can capture the temperature distribution of the crack area in real time, and generate a crack temperature change image of the tunnel by collecting temperature data at different time points for many times. The image will show the dynamic characteristics of the temperature change of the crack area over time. The crack temperature change image of the tunnel is analyzed at the pixel level to calculate the temperature difference between each adjacent pixel in the crack area. By counting these temperature differences, the crack temperature difference data of the tunnel is obtained. The temperature difference data can reflect the uniformity of heat distribution in the crack area, as well as the abnormal hot spots or cold spots that may exist. The temperature difference data is further analyzed to calculate the rate of temperature difference change in the crack area. The temperature difference change rate data can be obtained by differential analysis of the temperature difference at different time points, revealing the change in the speed of heat conduction and heat diffusion in the crack area, and generating temperature difference change rate data. The crack temperature change image is subjected to thermal balance analysis using the tunnel crack temperature difference data and temperature difference change rate data. The analysis process includes evaluating the state of heat conduction and heat diffusion in the crack area, determining whether the crack area is in a state of thermal equilibrium, or whether there is thermal imbalance. The analysis results can be achieved through numerical simulation or image analysis technology to generate tunnel crack thermal balance data. Thermal balance data is of great significance for judging the activity level and potential expansion risk of cracks. Based on the tunnel crack thermal balance data, the crack type area image is further analyzed using spectral analysis technology. Spectral analysis can reveal the optical properties of different materials in the crack area, such as reflectivity, absorptivity and emissivity. By analyzing these spectral characteristics, the material composition and degree of deterioration of the crack area can be judged, and the tunnel crack spectral analysis data can be generated. Using spectral analysis data, combined with thermal balance data, the types of cracks can be further classified and identified, such as thermal cracks, cold cracks or stress cracks. This process can improve the accuracy of crack detection and provide a scientific basis for subsequent tunnel maintenance and repair.

[0045] Preferably, step S34 includes the following steps: Step S341: performing Raman spectrum acquisition on the crack type area image according to the roadway crack thermal balance data based on the spectrum analysis technology to obtain the roadway crack spectrum map; Step S342: performing crack environment element analysis on the crack spectrum of the roadway to generate crack environment element data; identifying chemical composition of the crack type area data through the crack environment element data to generate crack environment chemical substance data; Step S343: analyzing the oxidation state of the crack material on the crack spectrum of the roadway to generate crack material oxidation state data; analyzing the oxidation degree of the crack type area data according to the crack material oxidation state data to generate crack oxidation degree data; Step S344: Analyze the changes in crack spectrum characteristics during the time period of the roadway crack spectrum diagram using the crack environmental chemical substance data and the crack oxidation degree data to generate roadway crack spectrum analysis data.

[0046] In an embodiment of the present invention, Raman spectrum acquisition is performed on the crack type area image using the tunnel crack thermal balance data based on spectral analysis technology. Raman spectroscopy technology irradiates the crack surface with laser and analyzes the frequency shift information in the scattered light to identify the molecular vibration mode of the crack material. This process can capture the spectral characteristics of different materials in the crack area and generate a tunnel crack spectrum. The spectrum shows the molecular structure information of the crack area and is the basis for subsequent analysis. A detailed environmental element analysis is performed on the tunnel crack spectrum. This includes identifying various elements present in the crack area, such as metal elements, non-metal elements and possible pollutants, through characteristic peaks in the spectrum. The analysis process can combine the Raman spectral database and matching algorithm to generate crack environmental element data. Using the crack environmental element data, the chemical composition of the crack type area data is identified. By comparing the spectral characteristics of known chemical substances, the chemical composition in the crack area is identified and crack environmental chemical substance data is generated. This process helps to understand the material composition of the crack area and possible external pollution sources. The oxidation state of the crack material is analyzed on the tunnel crack spectrum. By observing the specific oxidation characteristic peaks in the spectrum, the oxidation state of the material in the crack area can be determined, such as whether there is iron oxide, aluminum oxide, etc. The generated crack material oxidation state data can reflect the aging and deterioration of the crack material. According to the crack material oxidation state data, the crack type area data is analyzed for oxidation degree. Using quantitative analysis methods, the oxidation degree of different materials in the crack area is calculated, and crack oxidation degree data is generated. This data helps to assess the long-term stability of the crack and the risk of further expansion. Combined with the crack environmental chemical data and crack oxidation degree data, the time period change analysis of the spectral characteristics of the tunnel crack spectrum is performed. By comparing the spectral data of different periods, the changing trends of the material composition and oxidation state in the crack area are identified, and the tunnel crack spectrum analysis data is generated. This analysis can reveal the dynamic change characteristics of the crack and provide an important reference for predicting the development of the crack.

[0047] As an example of the present invention, refer to Figure 4 As shown, in this example, step S4 includes: Step S41: dividing the crack formation period based on the roadway crack heat balance data and the roadway crack spectrum analysis data to generate crack period data; Step S42: deduce and mark the fracture age of the fracture type area image using the fracture period data to generate fracture age deduction data; Step S43: Compare the fracture age deduction data with the preset fracture age range. When the fracture age deduction data is less than the preset fracture age range, the corresponding fracture type regional image is marked as a fracture new formation period image; when the fracture age deduction data is within the preset fracture age range, the corresponding fracture type regional image is marked as a fracture stable period image; when the fracture age deduction data is greater than the preset fracture age range, the corresponding fracture type regional image is marked as a fracture aging period image; Step S44: Perform crack age image discrimination on the crack type area annotated image. When it is confirmed that the crack type area image is a crack newly formed period image, high-frequency monitoring of the alley camera deployment data is performed. When it is confirmed that the crack type area image is a crack stable period image, medium-frequency monitoring of the alley camera deployment data is performed. When it is confirmed that the crack type area image is a crack aging period image, low-frequency monitoring of the alley camera deployment data is performed to execute dynamic monitoring operations for alley crack identification.

[0048] In an embodiment of the present invention, by analyzing the temperature change and spectral characteristics during the formation of the cracks based on the thermal balance data of the roadway cracks and the spectrum analysis data of the roadway cracks, the formation characteristics of the cracks at different stages are identified. Using these data, the formation period of the cracks is divided, such as the new formation period, the stable period and the aging period, and the corresponding crack period data are generated. These data will serve as the basis for subsequent crack age deduction and annotation. The formation age of the cracks is deduced by conducting an in-depth analysis of the crack period data. The specific method includes combining the thermal equilibrium state, spectral characteristics, and time series data of the cracks to calculate the formation time of the cracks, and annotating them on the crack type area image to generate crack age deduction data. This data can help identify the historical development of the cracks. The crack age deduction data is compared with the preset crack age range to determine the current stage of the crack: if the crack age deduction data is less than the preset crack age range, it is marked as a crack new formation period image. If the crack age deduction data is within the preset crack age range, it is marked as a crack stable period image. If the fracture age deduction data is greater than the preset fracture age range, it is marked as a fracture aging period image. When the fracture type area image is confirmed to be a fracture new formation period image, the tunnel camera deployment data is monitored at a high frequency. This means that in the stage of rapid development of fractures, more intensive monitoring is required to capture its changes. When the fracture type area image is confirmed to be a fracture stability period image, the tunnel camera deployment data is monitored at a medium frequency. The fracture is in a stable stage, and the monitoring frequency can be appropriately reduced, but regular observation is still required. When the fracture type area image is confirmed to be a fracture aging period image, the tunnel camera deployment data is monitored at a low frequency. The fracture enters the aging period and changes slowly, so the monitoring frequency can be further reduced.

[0049] Preferably, step S42 includes the following steps: Step S421: extracting crack period features from the crack period data using a preset crack period feature database to obtain crack period feature data; Step S422: dividing the data set of the crack period characteristic data to generate a model training set and a model test set; using the support vector machine algorithm to perform model training on the model training set to generate a crack age deduction pre-model; Step S423: The fracture age deduction pre-model is optimized and iterated through the model test set to generate a fracture age deduction model; the fracture type area image is imported into the fracture age deduction model to perform fracture age deduction and annotation to generate a fracture type area annotation image.

[0050] In an embodiment of the present invention, the crack period data is feature extracted through a preset crack period feature database. These features may include the thermal equilibrium state, spectral characteristics, extension direction and speed of the crack. By extracting these feature data, valuable input can be provided for subsequent model training. Finally, crack period feature data are obtained, which reflect the unique behavior patterns of cracks in different periods. The crack period feature data is divided into a model training set and a model test set. The model training set is used to establish the model, while the model test set is used to verify the performance and accuracy of the model. The model training set is trained using a support vector machine (SVM) algorithm to generate a crack age deduction pre-model. Support vector machine is a machine learning algorithm commonly used for classification and regression analysis, and is particularly suitable for processing complex high-dimensional data, such as crack period feature data. The crack age deduction pre-model is optimized and iterated through the model test set, and the parameters and structure of the model are adjusted to improve the accuracy of its prediction. Finally, an optimized crack age deduction model is generated. The crack type area image is imported into the optimized crack age deduction model, the crack age is deduced, and the results are annotated. In this way, annotated images of fracture type areas can be automatically generated, providing accurate references for subsequent monitoring and maintenance work.

[0051] In this specification, a tunnel crack identification system based on image recognition is provided, which is used to execute the above-mentioned tunnel crack identification method based on image recognition. The tunnel crack identification system based on image recognition includes: The surface crack area recognition module is used to obtain the tunnel structure data; perform regional camera deployment according to the tunnel structure data to obtain the tunnel camera deployment data; perform tunnel image acquisition according to the tunnel camera deployment data to generate a tunnel time series image set; perform tunnel morphology change analysis on the tunnel time series image set to generate tunnel time series change data; perform surface crack area recognition on the tunnel time series image set through the tunnel time series change data to generate a tunnel surface crack area image; The hidden crack area identification module is used to perform surface crack trend analysis on the crack area image on the roadway surface to generate surface crack texture trend data; perform depth map conversion based on the crack area image on the roadway surface to generate a crack area depth map on the roadway surface; perform crack ductility analysis on the crack area depth map on the roadway surface to generate surface crack ductility data; use the surface crack texture trend data and surface crack ductility data to confirm the hidden roadway crack area on the crack area image on the roadway surface to obtain a hidden roadway crack area image; The crack feature recognition module is used to integrate the tunnel hidden crack area image and the tunnel surface crack area image into a crack type area image to obtain a crack type area image; perform crack thermal balance analysis on the crack type area image to generate tunnel crack thermal balance data; perform crack spectrum analysis on the crack type area image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data; The monitoring frequency adjustment module is used to divide the crack formation period based on the tunnel crack thermal balance data and the tunnel crack spectral analysis data to generate crack period data; to deduce and mark the crack age of the crack type area image through the crack period data to generate crack age deduction data; and to adjust the monitoring frequency of the tunnel camera deployment data according to the crack age deduction data to perform dynamic monitoring operations for tunnel crack identification.

[0052] The beneficial effect of the present invention is that by acquiring the tunnel structure data and deploying cameras, the structural layout of the tunnel can be fully understood, ensuring that data collection covers the entire tunnel area, and ensuring the comprehensiveness and accuracy of subsequent analysis. Analyzing the tunnel time-series image set can monitor the changes in the tunnel morphology, identify potential surface crack areas, and provide key data support for the timely discovery and treatment of cracks. Through morphological change analysis and surface crack area identification, the crack area can be effectively distinguished, laying the foundation for subsequent crack feature analysis and evaluation. Analyzing the trend of surface cracks can understand the expansion direction of the cracks and provide valuable information for further crack management and repair. Through depth map conversion and crack ductility analysis, the width and depth of the cracks can be accurately measured, the expansibility of the cracks can be evaluated, and the degree of impact of the cracks on the tunnel structure can be evaluated. Combining crack trend data and ductility data to confirm the hidden crack area helps to discover those cracks that may not be easy to detect, so as to carry out maintenance in advance and avoid potential risks. Integrating the hidden crack area image with the surface crack area image provides a comprehensive crack type area image, which helps to better understand the distribution and properties of the cracks. Through the crack heat balance analysis, the thermal characteristics of the cracks can be understood, which helps to reveal the formation mechanism of the cracks and their impact on the tunnel structure. Spectral analysis of the cracks can detect the chemical composition and material properties in the cracks, provide an in-depth understanding of the causes of the cracks, and provide a scientific basis for formulating repair strategies. According to the crack heat balance data and spectral analysis data, the crack formation period can be divided, the age of the cracks can be determined, the evolution process of the cracks can be understood, and time-effective information can be provided for crack management. Through crack age deduction and labeling, the age status of different cracks can be accurately understood, so as to formulate corresponding maintenance strategies according to the actual situation of the cracks. According to the crack age deduction data, the monitoring frequency of the tunnel camera can be adjusted to achieve targeted dynamic monitoring, improve monitoring efficiency and effect, and ensure timely discovery and handling of new crack problems. Therefore, the present invention improves the accuracy and comprehensiveness of identification through dynamic change detection, hidden crack identification, crack classification and analysis, and crack age deduction.

[0053] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0054] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A tunnel crack identification method based on image recognition, characterized in that: The following steps are involved: Step S1: Acquire tunnel structure data; perform regional camera deployment according to the tunnel structure data to obtain tunnel camera deployment data; perform tunnel image acquisition according to the tunnel camera deployment data to generate a tunnel time-series image set; perform tunnel morphology change analysis on the tunnel time-series image set to generate tunnel time-series change data; perform surface crack area recognition on the tunnel time-series image set using the tunnel time-series change data to generate a tunnel surface crack area image; Step S2: performing surface crack trend analysis on the crack area image on the tunnel surface to generate surface crack texture trend data; Performing depth map conversion based on the crack area image on the roadway surface to generate a depth map of the crack area on the roadway surface; Conduct crack ductility analysis on the depth map of crack areas on the roadway surface to generate surface crack ductility data; The surface crack texture trend data and surface crack ductility data are used to confirm the hidden crack area of ​​the roadway surface crack area image, and the hidden crack area image of the roadway is obtained; Step S3: integrating the tunnel hidden crack region image and the tunnel surface crack region image into a crack type region image to obtain a crack type region image; performing crack thermal balance analysis on the crack type region image to generate tunnel crack thermal balance data; performing crack spectrum analysis on the crack type region image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data; Step S4: dividing the crack formation period based on the roadway crack heat balance data and the roadway crack spectrum analysis data to generate crack period data; performing crack age deduction and labeling on the crack type area image through the crack period data to generate crack age deduction data; The monitoring frequency of the tunnel camera deployment data is adjusted according to the crack age deduction data to perform dynamic monitoring operations for tunnel crack identification.

2. The method for identifying cracks in tunnels based on image recognition according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: using laser scanning to obtain tunnel structure data; Step S12: dividing the lane into regions according to the lane structure data to generate lane structure division data; deploying cameras based on the lane structure division data to obtain lane camera deployment data; Step S13: performing lane image acquisition according to the lane camera deployment data at a preset time interval to generate a lane time series image set; performing image preprocessing on the lane time series image set to generate a standard lane time series image set, wherein the image preprocessing includes image filtering, image brightness enhancement and image standardization; Step S14: Analyze the changes in lane morphology of the standard lane time series image set to generate lane time series change data; identify surface crack areas of the standard lane time series image set using the lane time series change data to generate lane surface crack area images.

3. The method for identifying cracks in tunnels based on image recognition according to claim 2, characterized in that: Step S14 includes the following steps: Step S141: performing feature point recognition on the standard lane time series image set to generate feature points of the standard lane time series image; using the feature points of the standard lane time series image to unify the image perspective of the standard lane time series image set to generate a unified perspective lane image set; Step S142: extracting multi-dimensional morphological features from the unified perspective lane image set to obtain lane multi-dimensional morphological feature data, wherein the multi-dimensional morphological feature extraction includes edge line extraction, surface texture extraction, and geometric contour extraction; calculating the morphological variation of the unified perspective lane image set according to the lane multi-dimensional morphological feature data to generate lane time series morphological variation data; Step S143: using the lane time series morphological change data to segment the lane image set with the uniform view angle into abnormal lane change regions, and generate abnormal lane change region images; performing crack feature matching on the abnormal lane change region images, and generating crack feature matching data; Step S144: annotating the surface crack area of ​​the abnormal tunnel change area image according to the crack feature matching data to generate a tunnel surface crack area image.

4. The method for identifying cracks in tunnels based on image recognition according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: extracting surface crack textures from the crack area image on the tunnel surface to obtain surface crack texture distribution data; performing texture orientation analysis on the surface crack texture distribution data to generate surface crack texture orientation data; performing crack orientation trend analysis on the surface crack texture distribution data according to the surface crack texture orientation data to generate surface crack texture orientation trend data; Step S22: performing depth map conversion based on the crack area image on the tunnel surface to generate a crack area depth map on the tunnel surface; measuring the crack width and crack depth on the crack area depth map on the tunnel surface to obtain surface crack width data and surface crack depth data; Step S23: performing crack ductility analysis based on the surface crack width data and the surface crack depth data to generate surface crack ductility data; Step S24: using the surface crack texture trend data and the surface crack ductility data to confirm the hidden roadway crack area of ​​the roadway surface crack area image, and obtain the hidden roadway crack area image.

5. The method for identifying cracks in tunnels based on image recognition according to claim 1, characterized in that: Step S24 includes the following steps: Step S241: performing crack segment fitting on the crack region image on the surface of the tunnel according to the surface crack ductility data to generate crack extension speed assessment data; Step S242: Perform crack texture line direction vector fitting on the crack area image on the surface of the tunnel according to the surface crack texture trend data to generate crack extension direction assessment data. Step S243: Performing implicit crack association analysis based on the crack extension speed assessment data and the crack extension direction assessment data to generate implicit crack association data; Step S244: Perform stress field simulation on the crack area image on the tunnel surface based on the hidden crack association data to generate a hidden crack area image of the tunnel.

6. The method for identifying cracks in tunnels based on image recognition according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: integrating the tunnel hidden crack region image and the tunnel surface crack region image into a crack type region image to obtain a crack type region image; collecting crack temperature changes in the crack type region image based on infrared thermal imaging technology to generate a tunnel crack temperature change image; Step S32: performing adjacent temperature difference calculation on the tunnel crack temperature change image to obtain the tunnel crack temperature difference data; performing temperature difference change rate calculation on the tunnel crack temperature difference data to obtain the temperature difference change rate data; Step S33: performing crack heat balance analysis on the crack temperature change image of the tunnel using the crack temperature difference data and the temperature difference change rate data to generate crack heat balance data of the tunnel; Step S34: Based on the spectral analysis technology, crack spectrum analysis is performed on the crack type area image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data.

7. The method for identifying cracks in tunnels based on image recognition according to claim 6, characterized in that: Step S34 includes the following steps: Step S341: performing Raman spectrum acquisition on the crack type area image according to the roadway crack thermal balance data based on the spectrum analysis technology to obtain the roadway crack spectrum map; Step S342: performing crack environment element analysis on the crack spectrum of the roadway to generate crack environment element data; identifying chemical composition of the crack type area data through the crack environment element data to generate crack environment chemical substance data; Step S343: analyzing the oxidation state of the crack material on the crack spectrum of the roadway to generate crack material oxidation state data; analyzing the oxidation degree of the crack type area data according to the crack material oxidation state data to generate crack oxidation degree data; Step S344: Analyze the changes in crack spectrum characteristics during the time period of the roadway crack spectrum diagram using the crack environmental chemical substance data and the crack oxidation degree data to generate roadway crack spectrum analysis data.

8. The method for identifying cracks in tunnels based on image recognition according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: dividing the crack formation period based on the roadway crack heat balance data and the roadway crack spectrum analysis data to generate crack period data; Step S42: deduce and mark the fracture age of the fracture type area image using the fracture period data to generate fracture age deduction data; Step S43: Compare the fracture age deduction data with the preset fracture age range. When the fracture age deduction data is less than the preset fracture age range, the corresponding fracture type regional image is marked as a fracture new formation period image; when the fracture age deduction data is within the preset fracture age range, the corresponding fracture type regional image is marked as a fracture stable period image; when the fracture age deduction data is greater than the preset fracture age range, the corresponding fracture type regional image is marked as a fracture aging period image; Step S44: Perform crack age image discrimination on the crack type area annotated image. When it is confirmed that the crack type area image is a crack newly formed period image, high-frequency monitoring of the alley camera deployment data is performed. When it is confirmed that the crack type area image is a crack stable period image, medium-frequency monitoring of the alley camera deployment data is performed. When it is confirmed that the crack type area image is a crack aging period image, low-frequency monitoring of the alley camera deployment data is performed to execute dynamic monitoring operations for alley crack identification.

9. The method for identifying cracks in tunnels based on image recognition according to claim 8, characterized in that: Step S42 includes the following steps: Step S421: extracting crack period features from the crack period data using a preset crack period feature database to obtain crack period feature data; Step S422: dividing the data set of the crack period characteristic data to generate a model training set and a model test set; using the support vector machine algorithm to perform model training on the model training set to generate a crack age deduction pre-model; Step S423: The fracture age deduction pre-model is optimized and iterated through the model test set to generate a fracture age deduction model; the fracture type area image is imported into the fracture age deduction model to perform fracture age deduction and annotation to generate a fracture type area annotation image.

10. A tunnel crack identification system based on image recognition, characterized in that: For executing the tunnel crack identification method based on image recognition as claimed in claim 1, the tunnel crack identification system based on image recognition comprises: The surface crack area recognition module is used to obtain the tunnel structure data; perform regional camera deployment according to the tunnel structure data to obtain the tunnel camera deployment data; perform tunnel image acquisition according to the tunnel camera deployment data to generate a tunnel time series image set; perform tunnel morphology change analysis on the tunnel time series image set to generate tunnel time series change data; perform surface crack area recognition on the tunnel time series image set through the tunnel time series change data to generate a tunnel surface crack area image; The hidden crack area identification module is used to perform surface crack trend analysis on the crack area image on the roadway surface to generate surface crack texture trend data; perform depth map conversion based on the crack area image on the roadway surface to generate a crack area depth map on the roadway surface; perform crack ductility analysis on the crack area depth map on the roadway surface to generate surface crack ductility data; use the surface crack texture trend data and surface crack ductility data to confirm the hidden roadway crack area on the crack area image on the roadway surface to obtain a hidden roadway crack area image; The crack feature recognition module is used to integrate the tunnel hidden crack area image and the tunnel surface crack area image into a crack type area image to obtain a crack type area image; perform crack thermal balance analysis on the crack type area image to generate tunnel crack thermal balance data; perform crack spectrum analysis on the crack type area image according to the tunnel crack thermal balance data to generate tunnel crack spectrum analysis data; The monitoring frequency adjustment module is used to divide the crack formation period based on the tunnel crack thermal balance data and the tunnel crack spectral analysis data to generate crack period data; to deduce and mark the crack age of the crack type area image through the crack period data to generate crack age deduction data; and to adjust the monitoring frequency of the tunnel camera deployment data according to the crack age deduction data to perform dynamic monitoring operations for tunnel crack identification.

Images