Intelligent identification method and system for wood structure crack of ancient building

Through multi-exposure imaging technology and machine learning algorithm combined with three-dimensional laser scanning, we can identify and measure cracks in ancient buildings, solve the problem of insufficient detection in the existing technology, realize efficient and accurate crack identification and evaluation, and provide a non-invasive approach The monitoring methods have improved the scientific nature of ancient buildings.

CN118968231AInactive Publication Date: 2024-11-15SHENYANG JIANZHU UNIVERSITY
View PDF 0 Cites 11 Cited by

Patent Information

Application Number
CN202410973463.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-19
Publication Date
2024-11-15
Estimated Expiration
Not applicable · inactive patent

AI Technical Summary

Technical Problem

The existing technology has insufficient immediate and precise damage assessment in the detection of cracks in ancient building wooden structures, resulting in the tiny cracks being ignored in the early stages until they develop into serious structural problems, and physical contact detection methods will damage the original structure.

Method used

Using multi-exposure imaging technology and machine learning algorithms, we can identify and measure cracks by generating crack visual enhancement maps, multi-scale feature extraction, residual connection refinement features, deep network processing and three-dimensional laser scanning technology, and generate cracks in combination with surface fitting algorithms. Structural stability evaluation results.

Benefits of technology

Improve the accuracy and efficiency of crack identification in ancient building wood structures, provide instant capture of initial damage, refine crack characteristics, ensure identification accuracy, and evaluate the impact of cracks on structural safety through three-dimensional structure and depth measurement, providing non-invasive monitoring means , improves the scientific nature of maintenance and repair work.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118968231A_ABST
    Figure CN118968231A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of historic building protection, in particular to an intelligent recognition method and system for historic building wood structure cracks, and the method comprises the following steps: capturing original image data of different exposure levels based on a multi-exposure imaging device, carrying out the optimization processing of original image details, carrying out the local contrast enhancement of a target crack region, and carrying out the recognition of the original image details; and generating a crack vision enhancement graph. According to the method, by combining the multi-exposure imaging technology and the machine learning algorithm, the precision and efficiency of ancient building wood structure crack recognition are improved, and by finely adjusting the image contrast and details, the visualization of micro cracks is improved, so that initial damage can be captured in real time; crack characteristics are refined, the accuracy of crack identification is enhanced by utilizing iterative optimization, and the three-dimensional structure and depth measurement of the crack are more accurate through point cloud data obtained by a three-dimensional laser scanning technology and a curved surface fitting algorithm, so that the influence of the crack on the structure safety is accurately evaluated.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of ancient building protection, and in particular to an intelligent identification method and system for cracks in wooden structures of ancient buildings. Background Art

[0002] The field of ancient building protection technology aims to maintain and repair buildings of historical value to ensure their structural integrity and cultural heritage. This field not only includes physical restoration work, such as the maintenance of stone, wood and other ancient materials, but also includes the use of modern technologies such as digital modeling, structural health monitoring and environmental impact assessment. With the assistance of science and technology, the damage of ancient buildings can be analyzed more accurately and more effective restoration strategies can be formulated. In addition, the protection of ancient buildings also involves the protection of the environment surrounding the buildings to ensure that the restoration work will not cause damage to the original cultural landscape.

[0003] Among them, the intelligent identification method of cracks in the wooden structure of ancient buildings is a technology that applies modern image processing and machine learning to the maintenance of the wooden structure of ancient buildings. It automatically identifies cracks and potential defects in the wooden structure through intelligent algorithms, so as to detect structural problems early and prevent safety risks. Its main purpose is to provide an efficient and accurate detection method without destroying the original building structure, helping protection personnel to better assess the health status of ancient buildings and perform necessary maintenance work.

[0004] Existing technologies for crack detection in wooden structures of ancient buildings rely on more traditional observations and simple technical means, which cannot provide immediate and accurate damage assessments. Traditional observation methods are inadequate in detecting tiny cracks, resulting in cracks being ignored in the early stages until they develop into serious structural problems. Existing technologies have obvious limitations in accurately measuring crack depth and three-dimensional structure, making it difficult to fully assess the potential impact of cracks on the overall structure, which not only increases maintenance costs, but also causes further damage to ancient buildings due to failure to intervene in time. In addition, detection methods that involve physical contact with ancient buildings will cause damage to the original structural materials. Summary of the invention

[0005] The purpose of the present invention is to solve the shortcomings of the prior art and to propose an intelligent identification method and system for cracks in wooden structures of ancient buildings.

[0006] In order to achieve the above object, the present invention adopts the following technical scheme: a method for intelligently identifying cracks in wooden structures of ancient buildings, comprising the following steps:

[0007] S1: Based on multi-exposure imaging equipment, capture the original image data with different exposure levels, optimize the details of the original image, and perform local contrast enhancement on the target crack area to generate a crack visual enhancement map;

[0008] S2: Based on the crack visual enhancement image, configure the input parameters of the YOLOv5 model, adjust the image size to match the model input format requirements, and perform multi-scale feature extraction on the adjusted image to generate a crack feature map;

[0009] S3: Based on the crack feature map, apply residual connection to refine crack features, iteratively optimize crack recognition accuracy, process crack features through a deep network layer, verify the accuracy of crack recognition, and generate a crack positioning frame;

[0010] S4: Based on the crack positioning frame, using 3D laser scanning technology, perform 3D scanning to capture geometric changes around the crack, obtain 3D point cloud data, the data includes 3D views and detail points of the crack, and generate a 3D crack model;

[0011] S5: Based on the three-dimensional crack model, analyzing the three-dimensional point cloud data, using a surface fitting algorithm to identify the position and size of the crack, and measuring the crack depth to generate a crack depth index;

[0012] S6: Based on the crack depth index, the crack data is integrated and analyzed to analyze the crack location, size and potential danger, and the impact of the cracks on the safety of the ancient building's wooden structure is evaluated to generate a structural stability assessment result.

[0013] As a further solution of the present invention, the crack visual enhancement map includes a detail enhancement area, a brightness enhancement area, and a contrast adjustment area; the crack feature map includes a crack edge definition result, a texture feature set, and a color difference area; the crack positioning frame includes a spatial coordinate system, a boundary length, and a boundary width; the three-dimensional crack model includes point cloud density, geometric shape description information, and a surface roughness index; the crack depth index includes the deepest point, the shallowest point, and a depth distribution map; the structural stability assessment result includes a risk level, an affected area size, and an emergency repair proposal.

[0014] As a further solution of the present invention, based on a multi-exposure imaging device, original image data with different exposure levels is captured, the details of the original image are optimized, and the local contrast of the target crack area is enhanced. The steps of generating a crack visual enhancement map are specifically as follows:

[0015] S101: Based on the multi-exposure imaging device, the enhanced image exposure technology is used to adjust the image brightness level, optimize the brightness difference in the crack area, and highlight the crack visualization features to obtain a brightness optimization map;

[0016] S102: Based on the brightness optimization map, the visual contrast between the crack and the surrounding wooden structure is optimized by modifying the brightness contrast parameter of the local area, and details are extracted to generate a visual contrast map;

[0017] S103: Based on the visual contrast image, sharpen the image edge, optimize the contrast of crack edge pixels, define the crack contour, and refine the image crack contour details to obtain a crack visual enhancement image.

[0018] As a further solution of the present invention, based on the crack visual enhancement map, the input parameters of the YOLOv5 model are configured, the image size is adjusted to match the model input format requirements, and multi-scale feature extraction is performed on the adjusted image to generate the crack feature map. Specifically, the steps are as follows:

[0019] S201: Based on the crack visual enhancement image, adjusting input parameters of the YOLOv5 model, including adjusting image size and resolution to match model input requirements, and determining compatibility and processing efficiency of the crack image to obtain a model-adapted image;

[0020] S202: Based on the model adaptation image, perform multi-scale crack feature extraction, including length and width information, adjust the scale level of the image, and analyze the image features of the difference level to generate a multi-scale feature map;

[0021] S203: Based on the multi-scale feature map, by fusing the crack information of differentiated scales, constructing an image feature description, and optimizing the model recognition accuracy, a crack feature map is generated.

[0022] As a further solution of the present invention, based on the crack feature map, residual connections are applied to refine crack features, the crack recognition accuracy is iteratively optimized, and the crack features are processed through a deep network layer to verify the accuracy of crack recognition. The steps of generating a crack positioning frame are specifically as follows:

[0023] S301: Based on the crack feature map, residual connection is used to optimize the crack feature stability, cross-layer connection is added, information flow is refined, and continuity of image features is adjusted to obtain a feature stability map;

[0024] S302: Based on the feature stability graph, perform multiple rounds of iterative optimization, adjust network weights and activation function parameters layer by layer, adjust recognition details of crack features, optimize crack recognition accuracy, and generate an optimized crack mapping graph;

[0025] S303: Based on the optimized crack map, use a deep learning network to verify and correct each identified crack feature, verify the positioning accuracy of the crack, and generate a crack positioning frame.

[0026] As a further solution of the present invention, the residual connection is according to the formula:

[0027]

[0028] Calculate the output features of each layer and generate a feature stability graph, where x represents the feature input to the residual block, W represents the weight matrix, b is the bias term, λ is the regularization parameter, and ‖W·x+b‖ is the Euclidean norm, which is used to calculate the standardized result of W·x+b. Represents the absolute value of the square root of the input feature x, e W·x+b refers to the exponential function used in the residual block.

[0029] As a further solution of the present invention, based on the crack positioning frame, three-dimensional scanning is performed using three-dimensional laser scanning technology to capture geometric changes around the crack and obtain three-dimensional point cloud data, the data including a three-dimensional view and detail points of the crack. The steps of generating a three-dimensional crack model are specifically as follows:

[0030] S401: Based on the crack positioning frame, using three-dimensional laser scanning technology, capturing details of the crack and surrounding space, collecting point cloud data, and optimizing the data spatial resolution to generate a point cloud data set;

[0031] S402: Based on the point cloud data set, analyze the spatial morphology in the point cloud, apply surface fitting technology, measure the width and depth of the crack, extract the three-dimensional shape of the crack, optimize the three-dimensional modeling process of the crack, and generate a crack space model;

[0032] S403: Based on the crack space model, the three-dimensional position and size data information of the cracks are integrated, and the data is used to evaluate the potential impact of the cracks on the safety of the wooden structure to generate a three-dimensional crack model.

[0033] As a further solution of the present invention, based on the three-dimensional crack model, the three-dimensional point cloud data is analyzed, the position and size of the crack are identified by using a surface fitting algorithm, and the crack depth is measured. The steps of generating the crack depth index are specifically as follows:

[0034] S501: Based on the three-dimensional crack model, adjust the density of the point cloud data, isolate the point cloud around the crack, optimize the contrast and clarity of the point cloud in the crack area, capture the crack space features, and generate a crack detail data set;

[0035] S502: Based on the crack detail data set, by calculating the morphological features of the point cloud data, measuring the three-dimensional shape, width and depth of the crack, and extracting the crack geometric features according to the detail analysis results to generate a crack geometric analysis diagram;

[0036] S503: Based on the crack geometry analysis diagram, data integration is performed to analyze the location and dimension information of the cracks, evaluate the potential impact of geometric characteristics on the integrity of the wood structure, and generate a crack depth index.

[0037] As a further solution of the present invention, based on the crack depth index, the crack data is integrated and analyzed to analyze the crack location, size and potential danger, and the impact of the cracks on the safety of the ancient building wooden structure is evaluated. The steps of generating the structural stability evaluation result are specifically as follows:

[0038] S601: Based on the crack depth index, the location, size and depth information of the cracks are integrated and compared with the standard index, the risk area is identified and marked, the potential structural hazards of the area are evaluated, and a risk identification record is generated;

[0039] S602: Based on the risk identification record, calculate and simulate the potential expansion path and development rate of the cracks, analyze the potential impact of factors on the long-term stability of the structure, and generate crack expansion analysis results;

[0040] S603: Based on the crack extension analysis results, the cumulative impact and potential hazard data of the cracks are integrated to evaluate the overall safety impact of the cracks on the wooden structure of the ancient building, and multiple rounds of verification are performed to determine the accuracy and reliability of the analysis results to generate a structural stability assessment result.

[0041] An intelligent recognition system for cracks in wooden structures of ancient buildings, the intelligent recognition system for cracks in wooden structures of ancient buildings is used to execute the above-mentioned intelligent recognition method for cracks in wooden structures of ancient buildings, and the system comprises:

[0042] The image acquisition module is based on a multi-exposure imaging device to capture raw image data with different exposure levels, optimize the details of the raw images, and perform local contrast enhancement on the target crack area to generate a crack visual enhancement map;

[0043] The feature extraction module uses the YOLOv5 model based on the crack visual enhancement image to adjust the image size to match the model input format requirements, and performs multi-scale feature extraction to generate a crack feature map;

[0044] The crack location module uses residual connection to refine crack features based on the crack feature map, verifies the accuracy of crack identification, and uses image recognition technology to determine the location and boundary of the crack and generate a crack location frame;

[0045] The 3D modeling module uses 3D laser scanning technology to capture geometric changes around the crack based on the crack positioning frame, obtains 3D point cloud data, including 3D views and detail points of the crack, and generates a 3D crack model;

[0046] The crack analysis and assessment module analyzes the three-dimensional point cloud data based on the three-dimensional crack model, uses a surface fitting algorithm to identify the location and size of the cracks, and integrates and analyzes the crack data to analyze the location, size and potential danger of the cracks, assess the impact of the cracks on the safety of the ancient building's wooden structure, and generate a structural stability assessment result.

[0047] Compared with the prior art, the advantages and positive effects of the present invention are:

[0048] In the present invention, by combining multi-exposure imaging technology and machine learning algorithm, the accuracy and efficiency of crack identification in the wooden structure of ancient buildings are improved. By fine-tuning the image contrast and details, the visualization of tiny cracks is improved, so that early damage can be captured immediately. The crack characteristics are refined and the accuracy of crack identification is enhanced by iterative optimization. The point cloud data and surface fitting algorithm obtained by three-dimensional laser scanning technology make the three-dimensional structure and depth measurement of cracks more accurate, thereby accurately evaluating their impact on structural safety, providing a non-invasive monitoring method, and greatly improving the scientific nature of the maintenance and restoration of wooden structures of ancient buildings. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0050] Figure 2 This is a detailed flow chart of S1 of the present invention;

[0051] Figure 3 This is a detailed flow chart of S2 of the present invention;

[0052] Figure 4 This is a detailed flow chart of S3 of the present invention;

[0053] Figure 5 This is a detailed flow chart of S4 of the present invention;

[0054] Figure 6 This is a detailed flow chart of S5 of the present invention;

[0055] Figure 7 This is a detailed flow chart of S6 of the present invention;

[0056] Figure 8 It is a system flow chart of the present invention. DETAILED DESCRIPTION

[0057] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0058] In the description of the present invention, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside" and the like indicate positions or positional relationships based on the positions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, in the description of the present invention, "multiple" means two or more, unless otherwise clearly and specifically defined.

[0059] Embodiment 1

[0060] See also Figure 1 The present invention provides a technical solution: an intelligent identification method for cracks in ancient building wooden structures, comprising the following steps:

[0061] S1: Based on multi-exposure imaging equipment, capture the original image data with different exposure levels, optimize the details of the original image, and perform local contrast enhancement on the target crack area to generate a crack visual enhancement map;

[0062] S2: Based on the crack visual enhancement image, configure the input parameters of the YOLOv5 model, adjust the image size to match the model input format requirements, and perform multi-scale feature extraction on the adjusted image to generate a crack feature map;

[0063] S3: Based on the crack feature map, residual connections are applied to refine crack features, the crack recognition accuracy is iteratively optimized, and crack features are processed through deep network layers to verify the accuracy of crack recognition and generate crack positioning frames;

[0064] S4: Based on the crack positioning frame, 3D laser scanning technology is used to perform 3D scanning to capture geometric changes around the crack, obtain 3D point cloud data, including 3D views and detail points of the crack, and generate a 3D crack model;

[0065] S5: Based on the 3D crack model, the 3D point cloud data is analyzed, and the surface fitting algorithm is used to identify the location and size of the cracks, and the crack depth is measured to generate a crack depth index;

[0066] S6: Based on the crack depth index, the crack data is integrated and analyzed to analyze the crack location, size and potential danger, and the impact of cracks on the safety of the ancient building’s wooden structure is evaluated to generate structural stability assessment results.

[0067] The crack visual enhancement map includes detail enhancement area, brightness enhancement area, and contrast adjustment area. The crack feature map includes crack edge definition results, texture feature set, and color difference area. The crack positioning frame includes spatial coordinate system, boundary length, and boundary width. The three-dimensional crack model includes point cloud density, geometric shape description information, and surface roughness index. The crack depth index includes the deepest point, the shallowest point, and depth distribution map. The structural stability assessment results include risk level, affected area size, and emergency repair proposals.

[0068] See also Figure 2 , based on multi-exposure imaging equipment, capture the original image data with different exposure levels, optimize the details of the original image, and perform local contrast enhancement on the target crack area. The specific steps for generating the crack visual enhancement map are as follows:

[0069] S101: Based on the multi-exposure imaging device, the enhanced image exposure technology is used to adjust the image brightness level, optimize the brightness difference in the crack area, and highlight the crack visualization features to obtain a brightness optimization map;

[0070] S101: Based on multi-exposure imaging equipment, first you need to set different exposure times to cover the brightness range from low to high, ensuring that each exposure level can capture the key details of the crack area. Use a brightness synthesis algorithm to integrate images with different exposures, and make adjustments specifically for insufficient lighting in the crack area. By enhancing the brightness of the area, the overall brightness contrast is optimized. This not only maintains the exposure balance of the image, but also enhances the brightness difference between the crack and the surrounding environment in the image, generating a brightness optimization map.

[0071] S102: Based on the brightness optimization map, the visual contrast between the crack and the surrounding wooden structure is optimized by modifying the brightness contrast parameters of the local area, and details are extracted to generate a visual contrast map;

[0072] S102: After obtaining the brightness optimization map, the brightness and contrast parameters of the local area are adjusted through software tools, focusing on the visual contrast between the crack and the surrounding wooden structure, making the crack more prominent by refining the brightness and contrast of the area, including using a high-pass filter to extract image details, especially the edges and internal texture of the crack, to generate a clearer visual contrast map.

[0073] S103: Based on the visual contrast image, the image edge is sharpened, the contrast of the crack edge pixels is optimized, the crack contour is defined, and the image crack contour details are refined to obtain a crack visual enhancement image;

[0074] S103: Based on the generated visual contrast map, the image edges are further sharpened, and the contrast of the crack edge pixels is optimized using an edge detection algorithm to clarify and refine the crack contour. In this way, the visualization features of the crack are further enhanced, making the crack more obvious in the entire image, and a crack visual enhancement map is obtained, focusing on improving the clarity and recognizability of the crack contour, providing a visual reference for subsequent analysis or repair work.

[0075] See also Figure 3 Based on the crack visual enhancement map, the input parameters of the YOLOv5 model are configured, the image size is adjusted to match the model input format requirements, and multi-scale feature extraction is performed on the adjusted image. The specific steps for generating the crack feature map are as follows:

[0076] S201: Based on the crack visual enhancement image, adjusting the input parameters of the YOLOv5 model, including adjusting the image size and resolution to match the model input requirements, and determining the compatibility and processing efficiency of the crack image to obtain a model-adapted image;

[0077] S201: Before using the YOLOv5 model, it is necessary to ensure that the compatibility and processing efficiency of the crack visual enhancement map meet the input requirements of the model. The specific operation is to adjust the size and resolution of the image. The adjustment is completed through a preset image processing script. The script automatically adjusts the image size according to the specified input size of the model. This process not only includes cropping or scaling the image, but also involves the conversion of color space to ensure that the format of the image data matches the model processing standard. The processing aims to maximize the retention of crack features while reducing the model processing time and improving the overall image processing process efficiency. Through adjustment, each image is converted into a model-adapted image, ready for further feature extraction and crack detection.

[0078] S202: Based on the model adaptation image, perform multi-scale crack feature extraction, including length and width information, adjust the scale level of the image, and analyze the image features of the difference level to generate a multi-scale feature map;

[0079] S202: After obtaining the model-adapted image, perform multi-scale crack feature extraction, adjust the scale level of the image, and use multi-scale processing technology to observe and record the changes in the length and width information of the cracks at different scales. By analyzing the cracks at different resolutions, the subtle changes and characteristics of the cracks can be captured more comprehensively, including using image pyramid technology to analyze the images at each scale level and extracting key feature information from each level, such as edge strength, direction, and texture changes. This multi-level analysis generates a multi-scale feature map containing rich crack feature information.

[0080] S203: Based on the multi-scale feature map, by fusing the crack information of differentiated scales, constructing an image feature description, optimizing the model recognition accuracy, and generating a crack feature map;

[0081] S203: Based on the generated multi-scale feature map, the crack information at different scales is merged through the feature fusion algorithm. This step mainly integrates the features generated at different scales to construct a comprehensive image feature description, which is crucial to improving the recognition accuracy of the model. During the feature fusion process, the features of each scale are weighted and integrated to ensure that the crack information from the most detailed to the broadest can be effectively recognized by the model. Through this method, a crack feature map is generated, which records the shape, size and performance of the cracks at different scales, thereby significantly optimizing the model's recognition accuracy of cracks.

[0082] See also Figure 4 , based on the crack feature map, residual connections are applied to refine crack features, the crack recognition accuracy is iteratively optimized, and crack features are processed through deep network layers to verify the accuracy of crack recognition. The specific steps for generating crack positioning frames are as follows:

[0083] S301: Based on the crack feature map, residual connection is used to optimize the crack feature stability, and cross-layer connection is added to refine the information flow and adjust the continuity of image features to obtain a feature stability map;

[0084] S301: Residual connection is used to optimize the stability of crack features. The core idea of ​​residual connection is to add cross-layer direct connection channels in deep networks, which can help information to be transmitted in the network and reduce the loss of information in the multi-layer transmission process. Specifically in crack detection, shallower image features (such as basic edge and texture information) are directly transmitted to deeper network levels through cross-layer connections. The connection not only improves the fluidity of features, but also enhances the continuity of image features in the entire network, ensuring that the features of the crack area are effectively expressed and strengthened at each layer of the network. In this way, the generated feature stability map not only retains the detailed information of the crack, but also increases the stability during feature transmission.

[0085] Residual connection, according to the formula:

[0086]

[0087] Calculate the output features of each layer and generate a feature stability graph, where x represents the features input to the residual block, W represents the weight matrix, which is used to adjust the influence of the convolutional layer in the residual block on the input features, b is the bias term, which is used to provide an offset for the nonlinear transformation inside the residual block, and λ is the regularization parameter, which controls the scaling ratio of the transformed features to ensure the stability of the features after transformation.

[0088] ‖W·x+b‖ is the Euclidean norm, which is used to calculate the normalized result of W·x+b to ensure the normalization of the feature conversion to maintain the uniformity of the features of each layer. Represents the absolute value of the square root of the input feature x, which is used to enhance the model's sensitivity to the details of the input feature. W·x+b refers to the exponential function used in the residual block.

[0089] Execution process:

[0090] By adjusting the convolutional layer parameters W and b in the residual block, the input feature x is appropriately transformed to obtain F(x). Next, the regularization parameter λ is applied to adjust the scaling of the features layer by layer to ensure the stability and continuity of the features when they are transmitted in multiple layers. At the same time, the transformation features ‖W·x+b‖ are standardized to ensure that the output features at different levels have similar amplitudes, thereby avoiding the problem of gradient vanishing or exploding. Enhance the model's ability to process input features in detail, improve the accuracy of crack detection, and generate feature stability diagrams.

[0091] S302: Based on the feature stability graph, perform multiple rounds of iterative optimization, adjust the network weights and activation function parameters layer by layer, adjust the recognition details of the crack features, optimize the crack recognition accuracy, and generate an optimized crack mapping map;

[0092] S302: After the feature stability diagram is generated, it enters a multi-round iterative optimization stage. In this stage, the network weights and activation function parameters are adjusted layer by layer to refine the recognition and extraction of crack features. Each round of iteration focuses on optimizing the parameters of a specific layer. By analyzing the results of the previous round of iterations, it is determined which parameters need to be adjusted to enhance the expressiveness of crack features and the accuracy of recognition. In this process, key adjustments include the threshold setting of the activation function and the weight update strategy to ensure that each iteration can accurately capture the slight changes in the cracks. After continuous iterative optimization, the network can more accurately identify and map crack features and generate an optimized crack mapping diagram.

[0093] S303: Based on the optimized crack mapping diagram, using a deep learning network, verifying and correcting each identified crack feature, verifying the accuracy of crack positioning, and generating a crack positioning frame;

[0094] S303: Based on the optimized crack mapping diagram, a deep learning network is used to further verify and correct each identified crack feature. This step is crucial to ensure the accuracy of crack positioning. Through the terminal correction of the network, each crack feature will be accurately analyzed to confirm whether its location information is consistent with the actual situation. In this process, attention is paid to the length, width and position errors of the crack, as well as the influence of the surrounding environment on the recognition results. After correction, a positioning frame is generated for each crack, accurately marking the position and shape of the crack, thereby completing the entire crack detection and recognition process.

[0095] See also Figure 5 , based on the crack positioning frame, 3D laser scanning technology is used to perform 3D scanning, capture the geometric changes around the crack, and obtain 3D point cloud data. The data includes 3D views and detail points of the crack. The specific steps for generating a 3D crack model are:

[0096] S401: Based on the crack positioning frame, use 3D laser scanning technology to capture the crack and surrounding space details, collect point cloud data, optimize the data spatial resolution, and generate a point cloud data set;

[0097] S401: Use 3D laser scanning technology to capture the details of the crack and its surrounding space based on the previously generated crack positioning frame. This step involves aiming the scanning device at the area indicated by the crack positioning frame and performing high-precision laser scanning. By adjusting the scanning parameters, such as the scanning angle and scanning frequency of the laser, point cloud data of sufficient density is obtained to ensure that the spatial characteristics of the crack and its surrounding structure can be recorded in detail. The collected point cloud data undergoes preliminary processing, such as denoising and data compression, to optimize the spatial resolution of the data and generate a detailed point cloud data set.

[0098] S402: Based on the point cloud data set, analyze the spatial morphology in the point cloud, apply the surface fitting technology, measure the width and depth of the crack, extract the three-dimensional shape of the crack, optimize the three-dimensional modeling process of the crack, and generate a crack space model;

[0099] S402: Based on the collected point cloud data set, perform in-depth analysis of the point cloud data, including using surface fitting technology to accurately measure and extract the width, depth and overall three-dimensional shape of the crack, and analyze the spatial morphological differences between the crack and non-crack areas in the point cloud through advanced computer vision algorithms, and accurately fit the geometric surface of the crack area. This step involves not only geometric calculations, but also hierarchical processing of the data to distinguish the complex structure inside the crack and the simple structure around it. In this way, a crack space model is generated, which can accurately reflect the three-dimensional characteristics and spatial position of the crack.

[0100] S403: Based on the crack space model, the three-dimensional position and size data information of the cracks are integrated, and the data is used to evaluate the potential impact of the cracks on the safety of the wood structure to generate a three-dimensional crack model;

[0101] S403: Based on the generated crack space model, a comprehensive assessment of the cracks is conducted, integrating the three-dimensional position and size data of the cracks. Structural analysis and safety assessment techniques are used to evaluate the potential impact of cracks on the safety of the wood structure. The risk of structural failure caused by cracks is predicted by simulating the behavior of cracks under different loads and environmental conditions. The data and analysis results together constitute the final three-dimensional crack model, which not only provides a detailed three-dimensional view and detail points of the cracks, but also contains important information about the impact of cracks on the safety of the entire structure. This will provide a scientific basis for subsequent repair work and structural reinforcement, and ensure the long-term stability and safety of the wood structure.

[0102] See also Figure 6 ,Based on the 3D crack model, the 3D point cloud data is analyzed, and the surface fitting algorithm is used to identify the location and size of the cracks and measure the crack depth. The specific steps for generating the crack depth index are as follows:

[0103] S501: Based on the three-dimensional crack model, the density of the point cloud data is adjusted, the point cloud around the crack is isolated, and the contrast and clarity of the point cloud in the crack area are optimized to capture the crack space features and generate a crack detail data set;

[0104] S501: Based on the three-dimensional crack model, the point cloud data is adjusted, including increasing the density of the point cloud so that the crack area can be observed more clearly. This involves using point cloud processing software to isolate the point cloud around the crack and optimize the contrast and clarity of the point cloud data in the crack area. Through this method, the spatial features of the crack, such as edges and surface textures, can be captured more accurately. The optimized point cloud data provides higher resolution visual information for the crack details, generating a crack detail data set containing detailed crack features.

[0105] S502: Based on the crack detail data set, the morphological features of the point cloud data are calculated to measure the three-dimensional shape, width and depth of the crack, and the crack geometric features are extracted according to the detail analysis results to generate a crack geometric analysis diagram;

[0106] S502: Based on the obtained crack detail data set, the three-dimensional shape, width and depth of the crack are measured by calculating the morphological features of the point cloud data. This step uses surface fitting technology to identify and quantify the geometric characteristics of the crack. By analyzing the morphological changes in the point cloud, the precise boundary and depth of the crack are identified. The analysis process not only includes simple size measurement, but also involves complex three-dimensional geometric feature extraction, such as the curvature and inclination of the crack. Through detailed geometric analysis, a three-dimensional geometric analysis diagram of the crack is generated.

[0107] S503: Based on the crack geometry analysis diagram, data integration is performed to analyze the location and dimension information of the cracks, evaluate the potential impact of geometric characteristics on the integrity of the wood structure, and generate a crack depth index;

[0108] S503: Based on the generated crack geometry analysis diagram, further data integration and analysis are performed, involving the summary of crack location and size information, and an assessment of how geometric characteristics affect the integrity of the wooden structure. Through a comprehensive analysis of crack depth and shape, the impact that the crack may have on the structure is assessed, thereby generating a crack depth index. The index provides important information about the severity of the crack and provides a scientific basis for repair and reinforcement to ensure the safety and stability of the structure.

[0109] See also Figure 7 , based on the crack depth index, the crack data is integrated and analyzed to analyze the crack location, size and potential danger, and the impact of cracks on the safety of the ancient building's wooden structure is evaluated. The specific steps for generating the structural stability assessment results are as follows:

[0110] S601: Based on the crack depth index, integrate the location, size and depth information of the cracks, compare with the standard index, identify and mark the risk area, evaluate the potential structural hazards in the area, and generate a risk identification record;

[0111] S601: Based on the obtained crack depth index, the specific location, size and depth information of the cracks are integrated. This step needs to be compared with the standard safety indicators in the industry to identify those crack areas that exceed the safety threshold. The areas are marked as high-risk areas. Through this comparison and marking, cracks that pose a threat to structural integrity can be identified. Subsequently, the risk areas are evaluated for potential structural hazards, considering how the location, size and depth of the cracks jointly affect the safety of the building, and generating a risk identification record that records the hazard rating and specific location information of each crack.

[0112] S602: Based on the risk identification records, calculate and simulate the potential expansion path and development rate of cracks, analyze the potential impact of factors on the long-term stability of the structure, and generate crack expansion analysis results;

[0113] S602: Based on the risk identification records, calculate and simulate the potential crack expansion path and development rate. This step involves complex structural analysis and uses computer simulation technology to predict the future expansion direction and speed of cracks. The factors analyzed include the morphology of existing cracks, the physical properties of wood, and environmental factors that have a potential impact on crack development. Through simulation, the trend of crack expansion can be predicted, the potential threat of crack development to the long-term stability of the building can be evaluated, and crack expansion analysis results can be generated.

[0114] S603: Based on the crack propagation analysis results, the cumulative impact and potential hazard data of cracks are integrated to evaluate the overall safety impact of cracks on the wooden structure of ancient buildings. Multiple rounds of verification are conducted to determine the accuracy and reliability of the analysis results and generate structural stability assessment results.

[0115] S603: Based on the results of the crack propagation analysis, a comprehensive assessment of the cumulative impact and potential hazards of the cracks is conducted. This step focuses on the overall structural safety analysis, assessing the combined impact of all identified cracks on the overall safety of the ancient building’s wooden structure. Multiple rounds of verification are conducted to determine the accuracy and reliability of the analysis results, ensuring that the data provided can accurately reflect the actual situation, and generating structural stability assessment results, which not only include a detailed analysis of the cracks, but also provide recommendations on how to repair or reinforce the structure to prevent future risks.

[0116] See also Figure 8 , an intelligent identification system for cracks in wooden structures of ancient buildings, the system comprising:

[0117] The image acquisition module is based on a multi-exposure imaging device to capture raw image data with different exposure levels, optimize the details of the raw images, and perform local contrast enhancement on the target crack area to generate a crack visual enhancement map;

[0118] The feature extraction module uses the YOLOv5 model based on the crack visual enhancement image to adjust the image size to match the model input format requirements, and performs multi-scale feature extraction to generate a crack feature map;

[0119] The crack location module uses residual connection to refine crack features based on the crack feature map, verifies the accuracy of crack identification, and uses image recognition technology to determine the location and boundary of the crack and generate a crack location frame;

[0120] The 3D modeling module uses 3D laser scanning technology based on the crack positioning frame to capture the geometric changes around the crack and obtain 3D point cloud data, which includes the 3D view and detail points of the crack, to generate a 3D crack model.

[0121] The crack analysis and assessment module is based on a three-dimensional crack model, analyzes three-dimensional point cloud data, uses a surface fitting algorithm to identify the location and size of cracks, and integrates and analyzes the crack data to analyze the location, size and potential danger of cracks, assess the impact of cracks on the safety of ancient building wooden structures, and generate structural stability assessment results.

[0122] The above are only preferred embodiments of the present invention and are not intended to limit the present invention in other forms. Any technician familiar with the profession may use the technical contents disclosed above to change or modify them into equivalent embodiments with equivalent changes and apply them to other fields. However, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention without departing from the technical solution of the present invention still falls within the protection scope of the technical solution of the present invention.

Claims

1. An intelligent method for identifying cracks in wooden structures of ancient buildings, characterized in that: The following steps are involved: Based on multi-exposure imaging equipment, the original image data with different exposure levels is captured, the details of the original image are optimized, and the local contrast of the target crack area is enhanced to generate a crack visual enhancement map; Based on the crack visual enhancement image, the input parameters of the YOLOv5 model are configured, the image size is adjusted to match the model input format requirements, and multi-scale feature extraction is performed on the adjusted image to generate a crack feature map; Based on the crack feature map, residual connections are applied to refine crack features, the crack recognition accuracy is iteratively optimized, and the crack features are processed through a deep network layer to verify the accuracy of crack recognition and generate a crack positioning frame; Based on the crack positioning frame, a three-dimensional laser scanning technology is used to perform a three-dimensional scan to capture geometric changes around the crack, obtain three-dimensional point cloud data, the data includes a three-dimensional view and detail points of the crack, and generate a three-dimensional crack model; Based on the three-dimensional crack model, the three-dimensional point cloud data is analyzed, and the position and size of the crack are identified by using a surface fitting algorithm, and the crack depth is measured to generate a crack depth index; Based on the crack depth index, the crack data is integrated and analyzed to analyze the crack location, size and potential danger, and the impact of cracks on the safety of the ancient building's wooden structure is evaluated to generate a structural stability assessment result.

2. The intelligent identification method for cracks in ancient building wood structures according to claim 1 is characterized in that: The crack visual enhancement map includes a detail enhancement area, a brightness enhancement area, and a contrast adjustment area. The crack feature map includes a crack edge definition result, a texture feature set, and a color difference area. The crack positioning frame includes a spatial coordinate system, a boundary length, and a boundary width. The three-dimensional crack model includes point cloud density, geometric shape description information, and a surface roughness index. The crack depth index includes the deepest point, the shallowest point, and a depth distribution map. The structural stability assessment result includes the risk level, the size of the affected area, and an emergency repair proposal.

3. The intelligent identification method for cracks in wooden structures of ancient buildings according to claim 1 is characterized in that: Based on the multi-exposure imaging device, the original image data with different exposure levels is captured, the details of the original image are optimized, and the local contrast of the target crack area is enhanced. The specific steps for generating the crack visual enhancement map are as follows: Based on multi-exposure imaging equipment, enhanced image exposure technology is used to adjust the image brightness level, optimize the brightness difference in the crack area, and highlight the crack visualization features to obtain a brightness optimization map; Based on the brightness optimization map, the visual contrast between the crack and the surrounding wooden structure is optimized by modifying the brightness contrast parameters of the local area, and details are extracted to generate a visual contrast map; Based on the visual contrast map, the image edge is sharpened, the contrast of crack edge pixels is optimized, the crack contour is defined, and the image crack contour details are refined to obtain a crack visual enhancement map.

4. The intelligent identification method for cracks in wooden structures of ancient buildings according to claim 1 is characterized in that: Based on the crack visual enhancement map, the input parameters of the YOLOv5 model are configured, the image size is adjusted to match the model input format requirements, and multi-scale feature extraction is performed on the adjusted image to generate the crack feature map. Specifically, the steps are as follows: Based on the crack visual enhancement image, adjusting input parameters of the YOLOv5 model, including adjusting image size and resolution to match model input requirements, and determining compatibility and processing efficiency of the crack image to obtain a model-adapted image; Based on the model adaptation image, perform multi-scale crack feature extraction, including length and width information, adjust the scale level of the image, and analyze the image features at the difference level to generate a multi-scale feature map; Based on the multi-scale feature map, by fusing the crack information of differentiated scales, an image feature description is constructed, and the model recognition accuracy is optimized to generate a crack feature map.

5. The intelligent identification method of cracks in ancient building wood structures according to claim 1 is characterized in that: Based on the crack feature map, residual connections are applied to refine crack features, the crack recognition accuracy is iteratively optimized, and the crack features are processed through a deep network layer to verify the accuracy of crack recognition. The steps of generating a crack positioning frame are specifically as follows: Based on the crack feature map, residual connections are used to optimize crack feature stability, and cross-layer connections are added to refine information flow and adjust the continuity of image features to obtain a feature stability map; Based on the feature stability graph, multiple rounds of iterative optimization are performed to adjust network weights and activation function parameters layer by layer, adjust the recognition details of crack features, optimize crack recognition accuracy, and generate an optimized crack mapping graph; Based on the optimized crack mapping diagram, a deep learning network is used to verify and correct each identified crack feature, verify the positioning accuracy of the crack, and generate a crack positioning frame.

6. The intelligent identification method for cracks in wooden structures of ancient buildings according to claim 5 is characterized in that: The residual connection is based on the formula: Calculate the output features of each layer and generate a feature stability graph, where x represents the feature input to the residual block, W represents the weight matrix, b is the bias term, λ is the regularization parameter, and ‖W·x+b‖ is the Euclidean norm, which is used to calculate the standardized result of W·x+b. Represents the absolute value of the square root of the input feature x, e W·x+b refers to the exponential function used in the residual block.

7. The intelligent identification method for cracks in ancient building wood structures according to claim 1 is characterized in that: Based on the crack positioning frame, a three-dimensional scan is performed using three-dimensional laser scanning technology to capture geometric changes around the crack and obtain three-dimensional point cloud data, which includes a three-dimensional view and detail points of the crack. The steps for generating a three-dimensional crack model are specifically as follows: Based on the crack positioning frame, three-dimensional laser scanning technology is used to capture the crack and surrounding space details, collect point cloud data, and optimize the data spatial resolution to generate a point cloud data set; Based on the point cloud data set, the spatial morphology in the point cloud is analyzed, the surface fitting technology is applied, the width and depth of the crack are measured, and the three-dimensional shape of the crack is extracted, the three-dimensional modeling process of the crack is optimized, and the crack space model is generated; Based on the crack space model, the three-dimensional position and size data information of the cracks are integrated, and the data is used to evaluate the potential impact of the cracks on the safety of the wooden structure to generate a three-dimensional crack model.

8. The intelligent identification method for cracks in ancient building wood structures according to claim 1 is characterized in that: Based on the three-dimensional crack model, the three-dimensional point cloud data is analyzed, the position and size of the crack are identified by using a surface fitting algorithm, and the crack depth is measured. The steps of generating the crack depth index are specifically as follows: Based on the three-dimensional crack model, the density of the point cloud data is adjusted, the point cloud around the crack is isolated, and the contrast and clarity of the point cloud in the crack area are optimized to capture the crack space features and generate a crack detail data set; Based on the crack detail data set, the morphological features of the point cloud data are calculated to measure the three-dimensional shape, width and depth of the crack, and the crack geometric features are extracted according to the detail analysis results to generate a crack geometric analysis diagram; Based on the crack geometry analysis diagram, data integration is performed to analyze the location and dimension information of the cracks, evaluate the potential impact of geometric characteristics on the integrity of the wood structure, and generate a crack depth index.

9. The intelligent identification method for cracks in ancient building wood structures according to claim 1 is characterized in that: Based on the crack depth index, the crack data is integrated and analyzed to analyze the crack location, size and potential danger, and the impact of the cracks on the safety of the ancient building's wooden structure is evaluated. The specific steps for generating the structural stability assessment results are as follows: Based on the crack depth index, the location, size and depth information of the cracks are integrated and compared with the standard index to identify and mark the risk area, evaluate the potential structural hazards of the area, and generate a risk identification record; Based on the risk identification records, calculate and simulate potential crack propagation paths and development rates, analyze the potential impact of factors on the long-term stability of the structure, and generate crack propagation analysis results; Based on the crack extension analysis results, the cumulative impact and potential hazard data of the cracks are integrated to evaluate the overall safety impact of the cracks on the wooden structure of the ancient building. Multiple rounds of verification are carried out to determine the accuracy and reliability of the analysis results, and generate structural stability assessment results.

10. An intelligent identification system for cracks in wooden structures of ancient buildings, characterized in that: According to any one of claims 1 to 9, the intelligent identification method for cracks in wooden structures of ancient buildings comprises: The image acquisition module is based on a multi-exposure imaging device to capture raw image data with different exposure levels, optimize the details of the raw images, and perform local contrast enhancement on the target crack area to generate a crack visual enhancement map; The feature extraction module uses the YOLOv5 model based on the crack visual enhancement image to adjust the image size to match the model input format requirements, and performs multi-scale feature extraction to generate a crack feature map; The crack location module uses residual connection to refine crack features based on the crack feature map, verifies the accuracy of crack identification, and uses image recognition technology to determine the location and boundary of the crack and generate a crack location frame; The 3D modeling module uses 3D laser scanning technology to capture geometric changes around the crack based on the crack positioning frame, obtains 3D point cloud data, including 3D views and detail points of the crack, and generates a 3D crack model; The crack analysis and assessment module analyzes the three-dimensional point cloud data based on the three-dimensional crack model, uses a surface fitting algorithm to identify the location and size of the cracks, and integrates and analyzes the crack data to analyze the location, size and potential danger of the cracks, assess the impact of the cracks on the safety of the ancient building's wooden structure, and generate a structural stability assessment result.

Citation Information

Cited By

  • Tunnel crack identification method and system based on refined point cloud data

    CN119323700A

  • Intelligent detection system for house outer wall cracks

    CN119394191A

  • Reconstruction method of historic building structure analysis model

    CN119760813A

  • Bridge crack detection method and system

    CN119763095A

  • Bridge crack detection method and system

    CN119763095B