A response field-based crack state evaluation method and system

Abstract

The application provides a crack state evaluation method and system based on a response field, image data of a structure surface to be analyzed is acquired and preprocessed, discrete crack segments are extracted, and structure attributes thereof are extracted; a crack structure graph between the crack segments is constructed based on the structure attributes; a structure response and a propagation mechanism are introduced based on the crack structure graph, a stable structure response field is formed through global information propagation; and crack dominant path extraction and crack state evaluation are realized based on response field distribution. The application can realize adaptive extraction of a crack dominant path and evaluation of a crack overall state.

Description

Technical Field

[0001] This application relates to the field of structural health monitoring and intelligent image analysis technology, specifically to a crack state assessment method and system based on response field. Background Technology

[0002] Cracks are common surface defects in engineering structures such as concrete structures, bridges, tunnels, and industrial components. Their occurrence is usually closely related to factors such as structural stress state, material properties, construction quality, and long-term environmental effects. The appearance of cracks is often an important early signal of structural damage or performance degradation. Therefore, timely and accurate detection and analysis of surface cracks are of great significance for ensuring the safe operation of engineering structures.

[0003] Traditional crack detection mainly relies on manual inspection, where professionals visually inspect the structural surface and record the location, length, and morphological characteristics of cracks. However, this method is not only inefficient, but the results are also easily affected by factors such as the experience level of the inspectors, the inspection environment, and subjective judgment, making it difficult to meet the application requirements of high efficiency and high consistency in large-scale infrastructure inspection tasks.

[0004] With the development of computer vision technology, automatic crack detection using image processing methods has gradually become a research hotspot. In existing technologies, one type of method is based on traditional image processing techniques, extracting crack regions through edge detection, thresholding, and morphological operations. For example, edge detection algorithms can be used to obtain regions of abrupt grayscale changes in an image, thereby extracting crack edge information; typical methods include the Canny operator. These methods have the advantages of simple implementation and high computational efficiency. However, in complex engineering environments, due to texture interference, lighting variations, and noise on the structural surface, crack edges often exhibit problems such as breakage, discontinuity, or even false detection, thus affecting the complete representation of the crack structure.

[0005] In recent years, with the development of deep learning technology, some studies have used convolutional neural networks to segment or detect crack images to improve crack recognition accuracy. These methods enhance crack detection capabilities in complex backgrounds to some extent through multi-scale feature extraction. However, because cracks typically exhibit slender, low-contrast, and spatially uneven structural features, the loss of subtle crack information during multi-layer downsampling in the network can easily occur, affecting the continuity and detail representation of the crack structure. Furthermore, deep learning methods usually rely on large amounts of labeled data, and the model's generalization ability still has certain limitations in different engineering scenarios.

[0006] Whether using traditional image processing methods or deep learning methods, the output is typically represented as a crack edge map or a crack segmentation map. In these results, the crack structure is often represented as multiple discrete crack segments. Due to factors such as image noise, occlusion, and detection errors, these crack segments usually exhibit varying degrees of breakage. If crack analysis is performed directly based on these discrete segments, it is difficult to accurately reconstruct the complete crack path, thus affecting the calculation of key parameters such as crack length, direction, and topology.

[0007] To address the problem of connecting crack segments, existing techniques typically employ morphological closing operations, region growing, or distance- and orientation-based rule-based methods. However, these methods mostly rely on local geometric features for judgment, such as simple conditions like centroid distance, orientation difference, or neighborhood continuity, lacking a unified modeling of the overall crack structure relationship. When multiple cracks are close to each other, intersect, or have complex branches, incorrect connections or omissions are prone to occur, thus reducing the accuracy and stability of crack structure recovery.

[0008] Furthermore, existing methods in crack analysis typically remain at the geometric level, describing crack length, width, or number, lacking a systematic characterization of the overall crack structure. Especially in complex crack scenarios, cracks often exhibit multi-branch propagation, local bending, and cross-connections; relying solely on local information is insufficient to reflect the structural importance and evolution trend of the crack on a global scale. Therefore, current technologies still have significant shortcomings in crack structure representation and state assessment.

[0009] In summary, existing crack detection and analysis methods mainly suffer from the following problems: First, crack results are presented in discrete fragments, making it difficult to form a continuous and complete global structural representation; second, the crack connection process relies on local rules and lacks global consistency constraints, making it prone to false connections; and third, there is a lack of a unified modeling mechanism for the overall topological relationship and importance of the crack structure, making it difficult to achieve an effective mapping from crack morphology to structural state.

[0010] Therefore, there is an urgent need for an analytical method that can characterize the consistency of crack structure globally and extract the dominant crack path and assess its state through structural information propagation, thereby improving the structural representation capability and assessment accuracy in complex crack scenarios. Summary of the Invention

[0011] The purpose of this application is to propose a crack state assessment method and system based on response field. By constructing a crack structure diagram and introducing a structural response propagation mechanism on the crack structure diagram, the method achieves adaptive extraction of the dominant crack path and assessment of the overall crack state.

[0012] In a first aspect, this application provides a crack state assessment method based on crack response field modeling, including: S1: Acquire image data of the surface of the structure to be analyzed and preprocess it; extract discrete crack segments from the preprocessed image; determine the structural properties of each crack segment; S2: Based on the structural properties of the crack segments, determine the candidate connection relationships between the crack segments; construct a crack structure graph with crack segments as nodes and candidate connection relationships as edges; S3: Based on the structural properties of the crack fragment, determine the initial response value of each node in the crack structure diagram and calculate the structural propagation weight between each node; based on the initial response value and the structural propagation weight, update the response value of each node through iterative propagation until convergence to a stable state to obtain the steady-state response field; the steady-state response field includes the steady-state response value of each node in the crack structure diagram. S4: Filter the nodes in the crack structure diagram based on the steady-state response field and response threshold to obtain the main crack candidate node set; based on the main crack candidate node set, extract the corresponding crack structure sub-graph from the crack structure diagram as the crack dominant path region; S5: Calculate the overall energy value of the crack structure based on the steady-state response value of each node in the candidate node set of the main crack; evaluate the state of the crack based on the overall energy value of the crack structure and the structural energy classification threshold. S6: Output the crack dominant path region and crack state assessment results.

[0013] In one possible implementation, the structural properties of the crack segment in S1 include: the centroid position of the crack segment; In S2, candidate connection relationships between crack segments are determined based on the structural properties of the crack segments, including: calculating the spatial adjacency between crack segments based on the centroid position of the crack segments; and determining candidate connection relationships based on the spatial adjacency between crack segments.

[0014] In one possible implementation, the structural properties of the crack segment in S1 also include: the length of the crack segment, local orientation information, and the average gradient intensity of the corresponding region; In S3, based on the structural properties of the crack fragments, the initial response values ​​of each node in the crack structure diagram are determined, and the structural propagation weights between each node are calculated, including: The initial response value of each node corresponding to each crack segment is calculated using the following formula: ; in, Indicates the first The path length of a crack segment This represents the maximum path length among all crack segments. Indicates the first The average gradient intensity of the region corresponding to each crack segment This represents the maximum average gradient intensity across all crack segments. and Represents the weighting coefficients, satisfying ; Based on the local orientation information of crack segments, the orientation difference between crack segments is calculated; the structural propagation weight between corresponding nodes is calculated according to the orientation difference between crack segments and the spatial adjacency, and the calculation formula is as follows: ; ; in, Represents a node With nodes Unnormalized structural propagation weights between them , Let be the set of edges in the crack structure diagram; Represents a node With nodes The degree of spatial adjacency between them; This indicates the directional difference between corresponding crack segments; and These represent the distance scale parameter and the orientation scale parameter, respectively. Represents a node With nodes Normalized structure propagation weights between them; Represents a node The set of adjacent nodes.

[0015] In one possible implementation, in S3, the response values ​​of each node are updated through iterative propagation, using the following formula: ; in, and They represent the first Second and third Node at the next iteration The response value; Indicates the first Node at the next iteration The response value.

[0016] In one possible implementation, in S3, convergence to a steady state includes: the change in the global response being less than a preset convergence threshold. The corresponding judgment formula is: ; in, This represents the set of nodes in the crack structure diagram, corresponding to the set of crack segments.

[0017] In one possible implementation, S4 contains the set of candidate nodes for the main crack. : ; in, The preset response threshold is set based on the statistical characteristics of the response value distribution; For nodes The steady-state response value.

[0018] In one possible implementation, the overall energy value of the crack structure in S5 is calculated using the following formula: ; in This represents the overall energy value of the crack structure. Represents a node The steady-state response value, Represents the set of candidate nodes for the main crack. The number of nodes in; In one possible implementation, in S5, the crack state is evaluated based on the overall energy value of the crack structure and the structural energy classification threshold, including: ; in, Indicates the crack condition category. , , These represent different stages of crack development, corresponding to minor, advanced, and severe conditions, respectively. and To preset energy grading thresholds, satisfying .

[0019] Secondly, this application provides a crack state assessment system based on crack response field modeling, comprising: The image preprocessing and crack extraction module is used to acquire image data of the surface of the structure to be analyzed and preprocess it, extract discrete crack segments from the preprocessed image, and determine the structural properties of each crack segment. The crack structure graph construction module is used to determine the candidate connection relationships between crack segments based on the structural properties of crack segments; and to construct a crack structure graph with crack segments as nodes and candidate connection relationships as edges. The response field construction module is used to determine the initial response value of each node in the crack structure diagram based on the structural properties of the crack fragment, and to calculate the structural propagation weight between each node; based on the initial response value and the structural propagation weight, the response value of each node is updated through iterative propagation until it converges to a steady state to obtain the steady-state response field; the steady-state response field includes the steady-state response value of each node in the crack structure diagram. The crack dominant path identification module is used to filter nodes in the crack structure diagram based on the steady-state response field and response threshold to obtain a set of candidate nodes for the main crack; based on the set of candidate nodes for the main crack, the corresponding crack structure sub-graph is extracted from the crack structure diagram as the crack dominant path region. The crack state assessment module is used to calculate the overall energy value of the cracked structure based on the steady-state response values ​​of each node in the candidate node set of the main crack; and to assess the crack state based on the overall energy value of the cracked structure and the structural energy classification threshold. The output module is used to output the crack dominant path region and crack state assessment results; The system is used to implement the above-described method.

[0020] Thirdly, this application provides an electronic device, including: a memory and a processor; The memory is used to store computer programs; The processor is used to invoke the computer program to execute the method described above.

[0021] Fourthly, this application provides a computer-readable storage medium storing a computer program that, when executed on an electronic device, causes the electronic device to perform the method described above.

[0022] Fifthly, this application provides a computer program product, including a computer program that, when run on an electronic device, causes the electronic device to perform the method described above.

[0023] The specific implementation methods of the second to fourth aspects of this application can refer to the implementation methods of the first aspect, and will not be elaborated here.

[0024] Through the above technical solution, this application realizes a complete mapping process from crack image to structural response field and then to crack state determination. Compared with the prior art, this application has the following beneficial effects: it can stably identify the dominant crack path in complex crack scenarios; it enhances the expressive ability of structurally consistent regions through a global propagation mechanism; and it elevates crack analysis from local geometric description to global structural modeling, thereby improving the accuracy and robustness of crack state assessment. Attached Figure Description

[0025] Figure 1 This is a flowchart of a method according to one embodiment of this application; Figure 2 This is a grayscale image of a crack image in one embodiment of this application; Figure 3 This is a schematic diagram of the crack structure in one embodiment of the present application; Figure 4 This is a schematic diagram illustrating the identification of the dominant crack path in one embodiment of this application. Detailed Implementation

[0026] To enable those skilled in the art to better understand the present application, the technical solution of the present application will be further described in detail below with reference to the embodiments and accompanying drawings.

[0027] This application relates to an automated analysis method for surface cracks in pavements and engineering structures, specifically disclosing a crack state assessment method based on crack structure response field modeling and propagation mechanism. This method integrates computer vision, crack structure diagram modeling, and information propagation theory. By extracting and modeling structural elements from crack images, it achieves the mapping and expression of cracks from low-level pixel representation to high-level structural semantics.

[0028] Furthermore, this application relates to structured analysis in image processing and pattern recognition, specifically crack skeleton extraction, geometric feature representation, crack structure map construction, and global consistency analysis techniques based on propagation models. By constructing a crack structure map and defining structural response functions and propagation mechanisms on it, a continuous structural response field is formed, thereby achieving crack main path identification and overall energy value modeling.

[0029] The method described in this application can be widely applied to road engineering inspection, bridge structure monitoring, concrete structure evaluation, and other engineering scenarios that require crack identification and condition determination. It is particularly suitable for stability analysis and evaluation under complex crack, multi-branch crack, and weak contrast crack environments.

[0030] Example 1: This embodiment provides a crack state assessment method based on response field, aiming to solve the problem of inaccurate crack state determination in complex crack scenarios due to multi-branch propagation, local noise interference, and irregular crack morphology. Figure 1 As shown, the method in this embodiment specifically includes: S1: Acquire image data of the surface of the structure to be analyzed and preprocess it; extract discrete crack segments from the preprocessed image; determine the structural properties of each crack segment; S2: Based on the structural properties of the crack segments, determine the candidate connection relationships between the crack segments; construct a crack structure graph with crack segments as nodes and candidate connection relationships as edges; S3: Based on the structural properties of the crack fragment, determine the initial response value of each node in the crack structure diagram and calculate the structural propagation weight between each node; based on the initial response value and the structural propagation weight, update the response value of each node through iterative propagation until convergence to a stable state to obtain the steady-state response field; the steady-state response field includes the steady-state response value of each node in the crack structure diagram. S4: Filter the nodes in the crack structure diagram based on the steady-state response field and response threshold to obtain the main crack candidate node set; based on the main crack candidate node set, extract the corresponding crack structure sub-graph from the crack structure diagram as the crack dominant path region; S5: Calculate the overall energy value of the crack structure based on the steady-state response value of each node in the candidate node set of the main crack; evaluate the state of the crack based on the overall energy value of the crack structure and the structural energy classification threshold. S6: Output the crack dominant path region and crack state assessment results.

[0031] The embodiments of this application are described in detail below in three parts: crack structure modeling, crack response field construction and crack dominant path identification based on the structure propagation model, and crack state assessment based on the structure response field. The effectiveness and stability of these embodiments are verified through specific experimental scenarios.

[0032] (a) Crack structure modeling: First, acquire image data of the surface of the structure to be analyzed. This image data can be collected using industrial cameras, mobile terminals, or drone platforms.

[0033] After obtaining the original image data, the original image data is standardized and preprocessed to improve the contrast of the crack area, thereby improving the identifiability and detectability of the crack area, as well as the stability of the structural representation in the subsequent structural modeling process.

[0034] The preprocessing process includes image grayscale conversion, noise filtering and suppression, and contrast enhancement to generate standardized image data.

[0035] Let the original input image be The preprocessing process can then be expressed as: ; That Represents the original input image. This represents the standardized image after preprocessing. This represents an image preprocessing function, which converts the original image into a standardized image with a uniform scale and grayscale range. In a preferred embodiment, this function may include operations such as Gaussian filtering, median filtering, and grayscale linear normalization.

[0036] This step yields a grayscale image of the crack, such as... Figure 2 As shown.

[0037] Subsequently, crack edge detection is performed on the preprocessed image to obtain the crack edge response image.

[0038] For example, gradient analysis is performed on the preprocessed normalized image using an edge detection operator to obtain the crack edge response image: ; in This represents the edge detection operator. This represents the crack edge response image. In this response image, high response areas correspond to locations in the image where grayscale changes are drastic, typically corresponding to crack edges or background texture boundaries.

[0039] Crack regions can be extracted using gradient-based edge detection algorithms or deep learning-based crack detection methods.

[0040] In some embodiments, by performing connected component analysis on the edge response map, continuous crack pixels are divided into several discrete crack segments, thereby forming a set of crack segments: ; in Indicates the first A crack segment, This indicates the total number of crack segments.

[0041] For each crack segment, its structural attribute information is extracted, such as the pixel set, centroid position, local orientation information, path length, and average gradient intensity of the corresponding region, to characterize the spatial structural characteristics and local image response characteristics of the crack segment.

[0042] Each crack segment consists of a spatially continuous set of pixels. A crack segment The corresponding set of pixels can be represented as: ; in Indicates the first A crack segment The first in The coordinates of each pixel. Indicates the first A crack segment The number of pixels it contains.

[0043] Furthermore, to characterize the local gradient intensity in the region corresponding to the crack segment, the first... Average gradient intensity of the region corresponding to each crack segment for: ; in, Indicates the image at pixel points The gray gradient vector at that point, This represents the corresponding gradient magnitude; The average gradient intensity of the region corresponding to the crack segment is used to reflect the local edge response characteristics of the crack segment.

[0044] To characterize the structural properties of crack segments, the path length of each crack segment is calculated: ; in Indicates the first The length of a crack segment, symbol This represents the Euclidean distance operation.

[0045] Furthermore, the principal direction model of the crack segment can be obtained through least squares fitting: ; in Indicates the first The slope of the fitted straight line for each crack segment. This represents the intercept parameter, from which the orientation angle of the crack segment can be obtained: ; in, Indicates the first The orientation angle of each crack segment, i.e., local orientation information.

[0046] Based on this, in order to characterize the positional distribution characteristics of crack segments in image space, the centroid position of each crack segment can be calculated. For the th... The centroid coordinates of a crack segment are defined as follows: ; in, Indicates the first The centroid coordinates of a crack segment and Let x and y represent the x-coordinate and y-coordinate of the centroid position of the crack segment in the image coordinate system (image space), respectively. The calculation formula is as follows: ; ; The above formula represents the average of the horizontal and vertical coordinates of all pixels within the crack segment to obtain the centroid coordinates of the crack segment in the image coordinate system.

[0047] After obtaining the crack segments and their structural properties, candidate connectivity relationships between the crack segments are further constructed. For any two crack segments... and The spatial distance between their centers of gravity is defined as:

[0048] ; in, Indicates a crack segment and The degree of spatial adjacency between two crack segments is the Euclidean distance between their centroids.

[0049] If the following conditions are met: ; Then it is considered a crack segment and There are candidate connections between them, where This represents a preset spatial neighborhood threshold used to control the connection range between crack segments. Its value can be set according to the image resolution or crack scale. For example, the value is 2% to 5% of the image diagonal length, or it can be directly set to a distance of 30 pixels.

[0050] Based on the above candidate connection relationships, a candidate connection set can be constructed: ; Furthermore, the set of crack fragments and the set of candidate connections are represented as a unified crack structure diagram: ; in This represents the set of nodes, corresponding to the set of crack segments. Indicates based on the set of candidate connections The resulting edge set is used to characterize the spatial connectivity between crack segments.

[0051] The above steps yield a crack structure diagram of the crack image, which characterizes the overall topological structure of the crack and provides a basis for subsequent structural response propagation. A schematic diagram of this crack structure diagram is shown below. Figure 3 As shown.

[0052] (II) Crack response field construction and crack dominant path identification based on structural propagation model

[0053] After constructing the crack structure diagram, it is necessary to further extract semantically meaningful substructures from the overall structure to achieve a hierarchical description of the crack system. To avoid the dependence of traditional path optimization methods on local connectivity, this application introduces a crack structure decomposition mechanism based on structural consistency propagation. By constructing a global structural response propagation model in the crack structure diagram, global information propagation calculations are performed on the nodes in the crack diagram to achieve adaptive identification of the dominant crack path.

[0054] Define the nodal structural response function on the crack structure diagram: ; in Represents a node The response value (structural response intensity) is a non-negative real number used to characterize the importance of the node in the overall cracked structure.

[0055] Furthermore, for any node in the crack structure diagram Define its initial structural response value as The initial structural response value is initialized based on the structural property information of the corresponding crack segment, specifically as follows: ; in, Indicates the first The path length of a crack segment This represents the maximum path length among all crack segments. Indicates the first The average gradient intensity of the region corresponding to each crack segment This represents the maximum average gradient intensity across all crack segments. and This represents the weighting coefficient, used to adjust the influence of path length features and gradient response features on the initial response value, and satisfies: ; By using the above-mentioned normalization initialization method, crack segments with longer path lengths or stronger edge response characteristics can obtain higher response values ​​in the early stages of propagation, thereby enhancing their dominant role in the subsequent structural propagation process.

[0056] Furthermore, for any connecting edge in the crack structure diagram Define nodes With nodes The unnormalized structure propagation weights between them are The specific calculation formula is as follows: ; in, Represents a node With nodes Spatial adjacency (spatial distance) between them; This represents the directional difference between corresponding crack segments. The value is the minimum of the absolute value of the difference between the orientation angles of the two crack segments after modulo π equivalent adjustment, in order to eliminate directional ambiguity. and These represent the distance scale parameter and the orientation scale parameter, respectively, used to control the degree of influence of spatial distance and orientation difference on the propagation weight. For example, It can be set as a neighborhood threshold. Half of, that is , Can be set to (Right now This ensures that segments with directional differences within this range maintain a high propagation weight. As can be seen from the above relationship, when the spatial distance between two nodes is small and their directional consistency is high, their corresponding propagation weights are larger, thereby enhancing the propagation ability of structural information between adjacent nodes.

[0057] Furthermore, to ensure the numerical stability of the propagation process, the unnormalized propagation weights are normalized to obtain the normalized structural propagation weights: ; in, Represents a node The set of adjacent nodes. Through the above normalization process, for any node... The sum of all its adjacency propagation weights satisfies: ; This ensures that the structural response has good numerical stability and convergence during the iterative propagation process.

[0058] Based on the above definition, a structural response propagation model is constructed on the crack structure diagram. This model updates the response values ​​of the nodes through iterative propagation, and the iterative update process is expressed as follows: ; in, and They represent the first Second and third Node at the next iteration The response value, Represents a node The set of adjacent nodes. This formula means that in each iteration, the response value of a node is obtained by weighted summation of the response values ​​of its adjacent nodes.

[0059] Through this propagation process, the structural response of the node will diffuse and redistribute throughout the crack structure diagram, causing the response value in the structurally consistent region to gradually increase, while the response value in the structurally discontinuous region to gradually decrease.

[0060] To ensure the termination and numerical stability of iterative calculations during propagation, a convergence determination mechanism is introduced.

[0061] In some embodiments, the propagation process is considered to have reached a stable state when the overall change in the node structure response between two adjacent iterations is less than a preset threshold. Specifically, the convergence criterion is defined as follows: ; in This represents the convergence threshold, which is a small positive real number used to control the termination precision of the propagation process. When the above condition is met, the iterative calculation stops, and the steady-state distribution of the node structure response is obtained. : ; This steady-state distribution reflects the global response intensity distribution of the crack structure, with regions of higher response values ​​corresponding to parts of the crack structure that exhibit strong connectivity and consistency. This steady-state response distribution also reflects the importance of each node in the crack diagram within the global structure, with nodes of higher response values ​​typically located on the dominant crack path.

[0062] Based on the steady-state response distribution, and according to the preset response threshold The nodal responses are filtered to obtain a set of candidate nodes for the main crack. : ; Among them, the response threshold This parameter is used to filter the set of nodes with high structural responses. The value can be set according to the statistical characteristics of the response value distribution, such as taking the mean or quantile of the response values ​​as the threshold.

[0063] The crack structure decomposition method based on the structure propagation model described above realizes the mapping process from local crack fragments to global structure representation. Compared with the traditional path optimization-based method, this method automatically enhances the structural consistency region through the global propagation mechanism, thereby enabling stable identification of the dominant crack path in complex crack, multi-branch structure and ring crack scenarios, and improving the robustness and accuracy of crack structure modeling.

[0064] Based on the candidate node set of the main crack Extract the corresponding crack structure sub-image This is used to characterize the region where the crack dominates. Thus, automatic identification of the crack dominance path is achieved, such as... Figure 4The diagram shown illustrates the identification of the dominant crack path.

[0065] (III) Crack state assessment based on structural response field

[0066] After completing the crack propagation model and obtaining the steady-state distribution of the nodal structural response. Subsequently, it is necessary to further map the global response characteristics of the cracked structure into crack state information with engineering significance. This application constructs an energy function for the cracked structure based on the structural response field, thereby achieving a unified characterization of the crack state.

[0067] Node-based steady-state structural response distribution Construct the energy function of the crack structure: ; in This represents the overall energy value of the crack structure. Represents a node The steady-state response value.

[0068] Based on the above definition, crack structure energy This reflects the distribution intensity of the crack structure response in the overall crack diagram, and its numerical value characterizes the connectivity and propagation of the crack structure globally. When the crack structure is relatively simple and has a clear dominant path, the structural response is mainly concentrated in a few continuous node regions, so the overall structural energy is relatively low. When the crack exhibits multi-branch propagation, complex connections, or large-scale propagation, the structural response will spread to more node regions, resulting in an increase in the overall structural energy.

[0069] Based on this, the structural energy space can be segmented to determine the crack state. In some embodiments, multiple energy grading thresholds can be preset to classify cracks into different state levels, such as minor cracks, developing cracks, and severe cracks.

[0070] In some embodiments, the crack state determination rule is expressed as: ; in, Indicates the crack condition category. , , These represent different stages of crack development. and To preset energy grading thresholds, satisfying Energy grading thresholds can be determined through statistical analysis of sample data, such as by dividing based on quantiles of structural energy distribution. For example, this can be achieved by analyzing the structural energy distribution of a large number of historical samples. Statistical analysis is conducted to determine, for example, setting the 33rd percentile of the energy value distribution as... The 66th percentile is set as This is based on the energy critical points that roughly correspond to the three states of minor, developing, and severe cracks, thereby enabling a reasonable distinction between different crack states.

[0071] This method enables a continuous mapping from the structural response field to the crack state, so that the crack evaluation process no longer relies on a simple combination of local features, but is based on the consistency of the global structure, thereby improving the analysis stability under complex crack scenarios.

[0072] The final output of this application is the crack condition assessment result and the corresponding crack dominant path information. The output result may also include the location of the crack dominant path, crack length, crack branching, and crack condition level.

[0073] Through the above embodiments, global response modeling and state assessment of crack structures can be achieved in complex crack scenarios. Compared with traditional methods based on local features or path optimization, this embodiment achieves globally consistent modeling through a response propagation mechanism. This enables the stable extraction of the dominant crack path under multi-branch crack and complex topology conditions, and improves the accuracy and robustness of crack state assessment.

[0074] Compared with traditional evaluation methods based on local geometric features or simple weighted indices, this method constructs a crack map and introduces a global response propagation mechanism to achieve adaptive redistribution of crack information in the global scope, thereby more stably depicting the overall development state of the crack.

[0075] Experimental verification: To verify the effectiveness and stability of the method in this embodiment in complex environments, a typical complex scenario—road surface crack images—was selected as the experimental object. Road surface cracks are usually accompanied by aggregate interference, uneven illumination, and random noise. Their structure exhibits multi-branching and irregular expansion characteristics, making it a typical scenario for testing the robustness of crack assessment methods.

[0076] The experimental dataset consists of 200 high-resolution images of pavement cracks, obtained through field collection. For each image, a crack state label is manually assigned as the baseline ground truth, categorized into three states: minor cracks, moderate cracks, and severe cracks. Random noise and local occlusion are introduced into some samples to create a more challenging testing environment. The dataset is divided into a parameter tuning set, a validation set, and a test set. The parameter tuning set is used to empirically determine hyperparameters (such as...). , , Reasonable values ​​for (etc.); the validation set is used to select the response threshold. The energy grading threshold is used; the test set is used to independently evaluate the final assessment performance. In this experiment, the method of this embodiment is compared with two typical methods: one is a traditional crack assessment method based on geometric feature weighting, and the other is a random forest classification method based on local statistical features. The traditional method obtains the crack score by linearly weighting features such as crack length, number of branches, and curvature change; the machine learning method takes local texture and geometric features as input and outputs the crack state category through a classification model.

[0077] To comprehensively evaluate the performance of different methods, this experiment uses classification accuracy and stability as evaluation criteria. Classification accuracy is defined as follows: ; variable Indicates classification accuracy. This indicates the number of crack image samples that were correctly classified. This represents the total number of crack image samples, and this metric is used to measure the overall accuracy of crack condition determination.

[0078] The stability index is further defined as follows: ; variable This indicates the stability of the evaluation results under perturbation conditions. Indicates the first The energy value corresponding to each original crack image sample Indicates the number after adding noise or blocking. The energy value corresponding to each crack image sample is used to measure the robustness of the method under external disturbances. Its value ranges from 0 to 1. The closer the value is to 1, the less sensitive the method is to disturbances.

[0079] The experimental results are shown in Table 1 below: Table 1 Comparison of Experimental Results

[0080] Experimental results show that traditional geometric feature-weighted methods are easily affected by local structural fluctuations in complex crack scenarios, resulting in low classification accuracy and significantly reduced stability under noise interference. Although random forest-based classification methods can improve classification accuracy, they still exhibit significant fluctuations in cases of structural fragmentation or occlusion due to their reliance on local features.

[0081] In contrast, the method provided in this embodiment achieves global integration of crack information through a structural response propagation mechanism, significantly enhancing the response of the dominant path and key structural regions, thereby focusing more on the overall morphological characteristics of the crack during condition assessment. Even under complex multi-branch crack and noise interference conditions, it maintains high classification accuracy and stability, verifying the effectiveness and robustness of this method in practical engineering scenarios.

[0082] Example 2: This embodiment provides a crack state assessment system based on crack response field modeling, including: The image preprocessing and crack extraction module is used to acquire image data of the surface of the structure to be analyzed and preprocess it, extract discrete crack segments from the preprocessed image, and determine the structural properties of each crack segment. The crack structure graph construction module is used to determine the candidate connection relationships between crack segments based on the structural properties of crack segments; and to construct a crack structure graph with crack segments as nodes and candidate connection relationships as edges. The response field construction module is used to determine the initial response value of each node in the crack structure diagram based on the structural properties of the crack fragment, and to calculate the structural propagation weight between each node; based on the initial response value and the structural propagation weight, the response value of each node is updated through iterative propagation until it converges to a steady state to obtain the steady-state response field; the steady-state response field includes the steady-state response value of each node in the crack structure diagram. The crack dominant path identification module is used to filter nodes in the crack structure diagram based on the steady-state response field and response threshold to obtain a set of candidate nodes for the main crack; based on the set of candidate nodes for the main crack, the corresponding crack structure sub-graph is extracted from the crack structure diagram as the crack dominant path region. The crack state assessment module is used to calculate the overall energy value of the cracked structure based on the steady-state response values ​​of each node in the candidate node set of the main crack; and to assess the crack state based on the overall energy value of the cracked structure and the structural energy classification threshold. The output module is used to output the crack dominant path and crack state assessment results; The system is used to implement the method described in Embodiment 1.

[0083] Example 3: This embodiment provides an electronic device, including: a memory and a processor; The memory is used to store computer programs; The processor is configured to invoke the computer program to execute the method as described in Embodiment 1.

[0084] Example 4: This embodiment provides a computer-readable storage medium storing a computer program. When the computer program is run on an electronic device, it causes the electronic device to perform the method described in Embodiment 1.

[0085] Example 5: This embodiment provides a computer program product, including a computer program that, when run on an electronic device, causes the electronic device to perform the method described in Embodiment 1.

[0086] The specific implementation of the system, electronic device, computer-readable storage medium, and computer program product provided in this application can be referred to the specific embodiments of the above methods, and will not be repeated here.

[0087] Obviously, those skilled in the art should understand that the various units or steps of this application described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby storing them in a storage device for execution by a computing device, or fabricating them separately as individual integrated circuit modules, or fabricating multiple modules or steps into a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0088] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A crack state assessment method based on crack response field modeling, characterized in that, include: S1: Acquire image data of the surface of the structure to be analyzed and preprocess it; extract discrete crack segments from the preprocessed image; determine the structural properties of each crack segment; S2: Based on the structural properties of the crack segments, determine the candidate connection relationships between the crack segments; construct a crack structure graph with crack segments as nodes and candidate connection relationships as edges; S3: Based on the structural properties of the crack fragment, determine the initial response value of each node in the crack structure diagram and calculate the structural propagation weight between each node; based on the initial response value and the structural propagation weight, update the response value of each node through iterative propagation until convergence to a stable state to obtain the steady-state response field; the steady-state response field includes the steady-state response value of each node in the crack structure diagram. S4: Based on the steady-state response field and response threshold, the nodes in the crack structure diagram are screened to obtain the set of candidate nodes for the main crack; Based on the set of candidate nodes for the main crack, the corresponding crack structure sub-graph is extracted from the crack structure graph and used as the dominant crack path region. S5: Calculate the overall energy value of the crack structure based on the steady-state response value of each node in the candidate node set of the main crack; evaluate the state of the crack based on the overall energy value of the crack structure and the structural energy classification threshold. S6: Output the crack dominant path region and crack state assessment results.

2. The method according to claim 1, characterized in that, In S1, the structural properties of the crack segment include: the position of the center of gravity of the crack segment; In S2, candidate connection relationships between crack segments are determined based on the structural properties of the crack segments, including: calculating the spatial adjacency between crack segments based on the centroid position of the crack segments; and determining candidate connection relationships based on the spatial adjacency between crack segments.

3. The method according to claim 2, characterized in that, In S1, the structural properties of the crack segment also include: the length of the crack segment, local orientation information, and the average gradient intensity of the corresponding region; In S3, based on the structural properties of the crack fragments, the initial response values ​​of each node in the crack structure diagram are determined, and the structural propagation weights between each node are calculated, including: The initial response value of each node corresponding to each crack segment is calculated using the following formula: ； in, Indicates the first The path length of a crack segment This represents the maximum path length among all crack segments. Indicates the first The average gradient intensity of the region corresponding to each crack segment This represents the maximum average gradient intensity across all crack segments. and Represents the weighting coefficients, satisfying ; Based on the local orientation information of crack segments, the orientation difference between crack segments is calculated; the structural propagation weight between corresponding nodes is calculated according to the orientation difference between crack segments and the spatial adjacency, and the calculation formula is as follows: ； ； in, Represents a node With nodes Unnormalized structural propagation weights between them , Let be the set of edges in the crack structure diagram; Represents a node With nodes The degree of spatial adjacency between them; This indicates the directional difference between corresponding crack segments; and These represent the distance scale parameter and the orientation scale parameter, respectively. Represents a node With nodes Normalized structure propagation weights between them; Represents a node The set of adjacent nodes.

4. The method according to claim 3, characterized in that, In S3, the response values ​​of each node are updated through iterative propagation, using the following formula: ； in, and They represent the first Second and third Node at the next iteration The response value; Indicates the first Node at the next iteration The response value.

5. The method according to claim 4, characterized in that, In S3, convergence to a steady state includes: the change in the global response being less than a preset convergence threshold. The corresponding judgment formula is: ； in, This represents the set of nodes in the crack structure diagram, corresponding to the set of crack segments.

6. The method according to claim 5, characterized in that, In S4, the set of candidate nodes for the main crack : ； in, The preset response threshold is set based on the statistical characteristics of the response value distribution; For nodes The steady-state response value.

7. The method according to claim 6, characterized in that, In S5, the overall energy value of the crack structure is calculated using the following formula: ； in This represents the overall energy value of the crack structure. Represents a node The steady-state response value, Represents the set of candidate nodes for the main crack. The number of nodes in the array.

8. The method according to claim 7, characterized in that, In S5, the crack state is evaluated based on the overall energy value of the crack structure and the structural energy classification threshold, including: ； in, Indicates the crack condition category. , , These represent different stages of crack development, corresponding to minor, advanced, and severe conditions, respectively. and To preset energy grading thresholds, satisfying .

9. A crack state assessment system based on crack response field modeling, characterized in that, include: The image preprocessing and crack extraction module is used to acquire image data of the surface of the structure to be analyzed and preprocess it, extract discrete crack segments from the preprocessed image, and determine the structural properties of each crack segment. The crack structure graph construction module is used to determine the candidate connection relationships between crack segments based on the structural properties of crack segments; and to construct a crack structure graph with crack segments as nodes and candidate connection relationships as edges. The response field construction module is used to determine the initial response value of each node in the crack structure diagram based on the structural properties of the crack fragment, and to calculate the structural propagation weight between each node. Based on the initial response value and the structural propagation weight, the response value of each node is updated through iterative propagation until it converges to a steady state, thus obtaining the steady-state response field; the steady-state response field includes the steady-state response value of each node in the crack structure diagram; The crack dominant path identification module is used to filter nodes in the crack structure diagram based on the steady-state response field and response threshold to obtain a set of candidate nodes for the main crack. Based on the set of candidate nodes for the main crack, the corresponding crack structure sub-graph is extracted from the crack structure graph and used as the dominant crack path region. The crack state assessment module is used to calculate the overall energy value of the cracked structure based on the steady-state response values ​​of each node in the candidate node set of the main crack. The state of the crack is assessed based on the overall energy value and structural energy classification threshold of the crack structure. The output module is used to output the crack dominant path region and crack state assessment results; The system is used to implement the method according to any one of claims 1 to 8.

Images