An image-based method and system for inspecting the rotation and closure of bridges.

Abstract

This invention provides an image-based method and system for bridge rotation and closure inspection, relating to the field of bridge rotation construction technology. The method includes acquiring a pre-closure final-state image of the bridge after rotation, including a top final-state image and a lateral final-state image; extracting structural geometric features of the closure region based on the pre-closure final-state image, and constructing a physical constraint reference system for the closure region using these features; performing feature mapping on the pre-closure final-state image under the physical constraint reference system to generate a closure feature vector of the bridge to be inspected; mapping the closure feature vector to a reference feature distribution space, calculating the degree of distribution deviation from the reference feature distribution, and obtaining the closure inspection result. This invention solves the problem that existing inspection methods cannot achieve automated closure inspection.

Description

Technical Field

[0001] This invention relates to the field of bridge rotation construction technology, and more specifically, to an image-based method and system for inspecting bridge rotation and closure. Background Technology

[0002] Bridge rotation construction is a common method for constructing long-span bridges, and the quality of its closure directly affects the structural safety and service performance of the bridge. Traditional closure inspection methods mainly rely on manual measurement, such as measuring the joint width with a steel ruler, measuring horizontal offset with a spirit level, and measuring longitudinal misalignment with a total station. These methods are not only labor-intensive but also difficult to obtain comprehensive three-dimensional structural information of the bridge closure, which is especially challenging for ultra-long-span bridges, making it difficult to guarantee accuracy and real-time performance.

[0003] With the development of computer vision and image processing technologies, some bridge rotation inspection methods have begun to utilize two-dimensional or three-dimensional images for joint width measurement, structural alignment analysis, and offset detection. However, these methods typically focus only on the characteristics of a single bridge or under a single construction condition, making it difficult to establish unified quantitative standards for bridges of different types, spans, and sizes. Furthermore, factors such as ambient temperature and thermal expansion and contraction significantly affect the minute deformations of bridge structures, and most existing methods do not consider the impact of temperature on the closure joint width and structural geometry, leading to biased inspection results. In addition, existing image processing methods mainly rely on geometric feature matching or image difference calculation, lacking statistical modeling and probabilistic evaluation methods for normal closure states. This makes it impossible to quantify the allowable deviation range for different bridges and to automatically determine whether the closure is abnormal. Especially during bridge rotation construction, even minute offsets can have a significant impact on the structural safety of the bridge, and existing methods cannot effectively identify these potential anomalies.

[0004] Therefore, there is an urgent need for a method that can comprehensively evaluate the closure status of bridges based on factors such as structural characteristics, type, span, size, and ambient temperature, to unify the inspection standards for different bridges, and to achieve automatic identification of abnormal closure status, thereby improving the scientificity, reliability, and safety of bridge rotation construction inspection. Summary of the Invention

[0005] The purpose of this invention is to provide an image-based method and system for inspecting the rotation and closure of bridges, thereby improving the aforementioned problems. To achieve this objective, the technical solution adopted by this invention is as follows:

[0006] Firstly, this application provides an image-based method for inspecting the rotation and closure of bridges, including:

[0007] Obtain the final state image of the bridge before closure after rotation, including the top final state image and the side final state image;

[0008] Based on the pre-closure final state image, the structural geometric features of the closure region are extracted, and a physical constraint reference system for the closure region is constructed using the structural geometric features.

[0009] Under the physical constraint reference frame, feature mapping is performed on the final state image before closure to generate the closure feature vector of the bridge to be tested;

[0010] The closure feature vector is mapped to the reference feature distribution space, and the degree of distribution deviation between the vector and the reference feature distribution is calculated to obtain the closure test result.

[0011] Secondly, this application also provides an image-based bridge rotation and closure inspection system, comprising:

[0012] The acquisition module is used to acquire the final state image of the bridge to be inspected before closure after rotation. The final state image before closure includes a top final state image and a side final state image.

[0013] An extraction and construction module is used to extract the structural geometric features of the closure region based on the pre-closure final state image, and to construct a physical constraint reference system for the closure region through the structural geometric features.

[0014] The mapping module is used to perform feature mapping on the pre-closure final state image under the physical constraint reference frame to generate the closure feature vector of the bridge to be inspected.

[0015] The verification module is used to map the closure feature vector to the reference feature distribution space, calculate the degree of distribution deviation between the vector and the reference feature distribution, and obtain the closure verification result.

[0016] The beneficial effects of this invention are as follows:

[0017] (1) This invention extracts structural geometric features from the top and lateral final state images of the bridge before closure after rotation, and combines this with information on bridge type, span, size, and closure environment temperature to construct a unified physical constraint reference system and embedding space, thereby achieving a standardized representation of the closure feature vector. This eliminates the influence of differences in shooting angle, equipment position, and bridge scale, making the closure features of different bridges and under different working conditions comparable, and improving the accuracy and robustness of feature description;

[0018] (2) This invention also constructs a reference feature distribution by embedding space and introduces structural scale adjustment terms and temperature coefficients to adjust the covariance, accurately characterizing the feature distribution of qualified bridge groups and calculating the degree of distribution deviation of the bridge under inspection. It can effectively identify abnormal states during the closure process, such as excessively narrow local joint width, vertical misalignment, or excessive planar offset, providing quantitative closure inspection results and providing a reliable basis for structural safety assessment.

[0019] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing embodiments of the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a schematic diagram of the image-based bridge rotation and closure inspection method described in this embodiment of the invention;

[0022] Figure 2 This is a schematic diagram of the image-based bridge rotation and closure inspection system described in an embodiment of the present invention. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the claimed invention, but merely to illustrate selected embodiments of the invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0024] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this invention, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0025] Example 1:

[0026] This embodiment provides an image-based method for inspecting the rotation and closure of bridges.

[0027] It should be noted that the closure inspection in this embodiment is conducted after the bridge rotation but before closure. Specifically, after the rotation is completed and the bridge reaches the designed angle, although the bridge structure is in its designed spatial position, the structures at both ends have not yet formed an integrated load-bearing system and are still in a temporary structural state before closure. During this stage, a certain closure gap is usually maintained between the structures on both sides of the main beam for subsequent closure operations such as wet joint construction, steel structure welding, or tensioning and locking. Once closure is completed, the structural stress state will change from a temporary state to a permanent state. If there are problems such as planar misalignment, vertical misalignment, or abnormal gap width before closure, the closure operation itself may introduce significant additional internal forces, even leading to an uncorrectable structural alignment. Therefore, systematically inspecting the final closure state (pre-closure state) after the bridge rotation and before closure is a crucial safety control step in bridge rotation construction. Through the closure inspection, it is possible to determine whether the current rotation position meets the design-allowed closure geometry conditions and to identify in advance the risks that may lead to forced closure, eccentric closure, or asymmetrical stress.

[0028] See Figure 1 The figure shows that the method includes steps S1, S2, S3 and S4.

[0029] Step S1: Obtain the final state image of the bridge before closure after rotation, including the top final state image and the side final state image;

[0030] In this step, the final state of the bridge before closure directly determines the closure quality. At this time, the structure has not yet formed an integral load-bearing system, which is the last window for deviation correction. Traditional single-view images cannot fully reflect the spatial characteristics of the closure area. Therefore, top and side dual-view acquisition is adopted. The top final state image is used to depict the projection geometry of the closure area in the bridge deck plane, which is suitable for analyzing the plane position, orientation, and alignment of the closure joint and the main beam axis. The side final state image is used to depict the structural relationship of the closure area in the vertical direction, which is suitable for analyzing the actual joint width, vertical misalignment, and elevation continuity of the upper and lower structures.

[0031] By introducing the aforementioned dual-view images before closure, a comprehensive assessment of the current rotation position can be conducted without physical closure, providing a basis for whether to carry out closure and closure adjustments, thereby reducing the possibility of structural errors or safety risks caused by closure deviations.

[0032] Step S2: Extract the structural geometric features of the closure region based on the pre-closure final state image, and construct a physical constraint reference system for the closure region using the structural geometric features;

[0033] In step S2, the closure region includes a top closure region and a side closure region. Extracting the structural geometric features of the closure region based on the pre-closure final state image includes:

[0034] Step S21: Based on the pre-closure final state image, locate the closure area of ​​the bridge rotation;

[0035] In this step, clearly defining the closure area in advance can provide spatial constraints and specificity for subsequent geometric quantity extraction.

[0036] Step S22: Extract the top physical quantities within the top closure area, including the top projected joint width curve, the set of planar edge points of the top closure joint, and the plane centerline of the main beam axis;

[0037] In this step, edge detection and contour fitting are used to obtain the planar edge point set of the top closure joint, as well as the left and right edge curves of the top closure joint in the bridge deck plane. The projected width of the top closure joint in the bridge deck plane is calculated using the left and right edge curves, thus obtaining the top projected joint width curve. Then, the edge point sets on both sides of the main beam are obtained, and the centerline is taken to obtain the centerline of the main beam axis plane.

[0038] Step S23: Extract lateral physical quantities within the lateral closure area, including the actual vertical joint width curve and the boundary elevation point set of the lateral closure joint;

[0039] In this step, the actual vertical joint width curve represents the vertical width variation of the lateral closure joint, reflecting dangerous conditions such as upper closure and lower compression. The boundary elevation point set of the lateral closure joint refers to a set of elevation points obtained by discretely sampling the structural boundary lines corresponding to the closure positions of the left and right bridges along the length of the lateral closure joint within the lateral closure area. By analyzing the elevation difference of the elevation point sets on both sides at positions with the same length parameter, the alignment status of the structures on both sides of the lateral closure joint in the vertical direction can be characterized, thereby identifying vertical misalignment and its spatial variation characteristics.

[0040] Step S24: Calculate the top geometric statistical characteristics through top physical quantities. The top geometric statistical characteristics include the lateral offset of the closure joint centerline in the plane, the plane deflection angle of the main beam axis, and the plane projection deviation index.

[0041] In this step, the center line of the closure joint is calculated by using the left and right edge curves within the plane of the bridge deck at the top closure joint. The center line of the closure joint represents the overall position of the top closure joint.

[0042] Lateral offset is used to characterize the asymmetric offset (translation error) of the closure joint centerline relative to the main beam axis in the bridge deck plane. The planar deflection angle of the main beam axis is used to characterize the angular error (rotation error) of the main beam axis deviating from the perpendicular state of the closure joint centerline in the bridge deck plane. The planar projection deviation index is used to characterize the degree of local geometric fluctuation (deformation error) of the closure joint centerline along the length of the closure joint in the bridge deck plane. Specifically, a straight line is fitted to the closure joint centerline in the bridge deck plane, the normal residual of each sampling point on the closure joint centerline relative to the fitted reference straight line is calculated, and the normal residual is statistically calculated to obtain the planar projection deviation index.

[0043] Step S25: Calculate the lateral geometric statistical characteristics through lateral physical quantities. The lateral geometric statistical characteristics include the actual top seam width, the actual bottom seam width, the seam width variation gradient, the vertical misalignment, and the vertical non-uniformity index.

[0044] In this step, the actual seam width at the top and bottom can reflect the risk of premature contact at one end and compression or obstruction at the other end. The seam width variation gradient can reflect the trend of the lateral closure seam along the vertical direction.

[0045] Vertical misalignment reflects the state of misalignment, such as left-high-right or right-high-left, i.e., the overall height difference. Specifically, along the vertical direction of the lateral closure joint (e.g., from the bridge deck to the bottom of the web), multiple corresponding height positions are selected. At each height position, the elevations of the left and right structures are extracted. Then, the single-point height difference between the left and right structure elevations at each height position is calculated. By averaging all single-point height differences, the vertical misalignment is obtained.

[0046] The vertical non-uniformity index reflects local fluctuations, warping, and torsion. Specifically, for the lateral closure joint along the vertical direction, after eliminating the influence of the overall vertical misalignment, the height differences between the left and right structures at multiple corresponding positions are obtained. After eliminating the overall height difference trend, the dispersion of the remaining height difference is statistically analyzed to obtain the vertical non-uniformity index.

[0047] Step S26: Structural geometric features are constructed using top and lateral geometric statistical features.

[0048] In step S2, constructing the physical constraint reference system for the closure region using the structural geometric features includes:

[0049] Step S27: Based on the center line of the main beam axis plane, determine the main structure reference axis, which serves as the longitudinal reference axis of the physical constraint reference system;

[0050] In this step, the extracted and statistically analyzed structural geometric features are uniformly mapped to a physical constraint reference system that is consistent with the bridge's own structure and independent of the imaging posture. This can eliminate differences in shooting angle and equipment position, making the features of different bridges and under different working conditions comparable.

[0051] Since the axis of the main beam is the most stable and structurally meaningful direction of the bridge, using it as a longitudinal reference can ensure that the coordinates are consistent with the actual structure of the bridge and are not affected by the rotation of the camera.

[0052] Specifically, the centerline of the main beam axis in the plane can be represented as a set of two-dimensional points. ,pass Perform a line fitting operation, use the axis corresponding to the fitted line as the reference axis of the main structure, and obtain the direction vector of the reference axis of the main structure. .

[0053] Step S28: Based on the main structure reference axis, determine a planar constraint reference axis orthogonal to the main structure reference axis in the bridge deck plane determined by the top closure area;

[0054] In this step, within the bridge deck plane, a direction vector orthogonal to the main structure reference axis is constructed. This yields a planar constraint reference axis orthogonal to the main structure reference axis.

[0055] Step S29: Determine the vertical physical constraint axis using the set of boundary elevation points of the vertical misalignment and the lateral closure joint;

[0056] The vertical physical constraint axis represents the direction in which, in the final state before closure, continued relative displacement within the closure area will first trigger structural contact and constraint. This direction truly reflects the vertical constraint characteristics during the closure process. For example, when the left bridge is higher than the right bridge, at the moment of closure, the left bridge will first make contact at the top or at a certain oblique position. After contact, the left bridge cannot continue to move downwards, and the right bridge cannot continue to rise. Therefore, the actual force direction forms a constraint along the direction of left-high and right-low, meaning this direction is not strictly vertical, but rather a slightly inclined vertical dominant direction along the misalignment direction from the left bridge to the right bridge. If further squeezing continues along this axis, the structure will experience contact, obstruction, and abrupt changes in force.

[0057] In this step, the vertical misalignment not only reflects the overall height difference between the left and right structures, but also induces lateral contact and restricted behavior during the closure process. Therefore, in order to characterize the true vertical constraint direction within the closure area, it is necessary to comprehensively consider the vertical misalignment and its coupling effect in the spatial distribution of the lateral closure joint.

[0058] Specifically, this is combined with the normal direction of the lateral closure joint in space. (Pointing to the direction from the left bridge to the right bridge) and the vertical misalignment, constructing the dominant vector of vertical misalignment. :

[0059] ;

[0060] In the formula, Indicates the vertical misalignment. Represents the global vertical unit vector. The lateral coupling coefficient, obtained by fitting the boundary elevation point set of the lateral closure joint, is used to reflect the lateral offset trend accompanying the vertical misalignment.

[0061] For the dominant vector of vertical misalignment After normalization, the direction vector of the vertical physical constraint axis is obtained. .

[0062] Step S210: Using the main structure reference axis, the planar constraint reference axis, and the vertical physical constraint axis as coordinate references, construct a physical constraint reference system for the closure region.

[0063] Understandably, different bridges have different structural parameters and shooting angles, and directly comparing the features of the original images can lead to biased judgments. The physical constraint reference system uses the bridge's own structure as a benchmark, rather than a fixed coordinate system, which can eliminate systematic errors caused by shooting posture and equipment position. The vertical physical constraint axis takes into account the coupling relationship between misalignment and elevation distribution, which is closer to the structural stress constraint logic in actual construction, providing an accurate coordinate benchmark for subsequent feature mapping.

[0064] Step S3: Under the physical constraint reference frame, perform feature mapping on the final state image before closure to generate the closure feature vector of the bridge to be inspected;

[0065] Step S3 includes:

[0066] Step S31: Map the pixel coordinates in the top final state image and the side final state image to a unified physical constraint reference system coordinate system to obtain the feature point set of the closure region;

[0067] In this step, through , and A coordinate transformation is performed to obtain the set of feature points in the closure area.

[0068] Step S32: Project the geometric features of the structure based on the feature point set of the closure region to obtain the three-dimensional structural quantity;

[0069] In this step, the structural geometric features are not extracted again. Instead, based on the feature points corresponding to the structural geometric features obtained in step S2, the coordinates of the structural geometric features are re-expressed and projected under the physical constraint reference system based on the feature point set of the closure region. This transforms the structural geometric features, which were originally represented in two-dimensional or local statistical form, into three-dimensional structural quantities.

[0070] Step S33: After tagging and encoding the three-dimensional structural quantities in conjunction with the bridge type, span grade, bridge size and ambient temperature at the time of closure of the bridge to be inspected, a closure feature vector of the bridge to be inspected is generated.

[0071] Step S4: Map the closure feature vector to the reference feature distribution space, calculate the degree of distribution deviation between the vector and the reference feature distribution, and obtain the closure test result.

[0072] Step S4 includes:

[0073] Step S41: Map the closure feature vectors of the verified bridge to the embedding space, and construct a reference feature distribution space based on the embedding space;

[0074] Step S41 includes:

[0075] Step S411: Obtain the closure feature vector of the verified bridge;

[0076] In this step, the closure feature vectors of multiple inspected bridges that have passed inspection are obtained, including multiple bridges that are on the verge of passing inspection.

[0077] Step S412: Map the closure feature vectors of the tested bridges to the embedding space to obtain the embedding representation of each tested bridge;

[0078] By using an embedding mapping function (such as a multilayer perceptron MLP or a fully connected network) to perform feature mapping on the closure feature vector of each tested bridge, the corresponding embedded representation of the tested bridge is obtained. This low-dimensional embedding space can capture the similarity between structure, scale and environmental conditions, and reduce the noise influence between high-dimensional features.

[0079] Step S413: Cluster the embedded representations of the verified bridges to obtain multiple embedded clusters;

[0080] In this step, the embedding representations in the embedding space are clustered using K-means or Gaussian Mixture Model (GMM) to obtain... Embedded clusters: ,in, Indicates the first An embedding cluster, , This represents the total number of embedding clusters. Embedding clusters can capture the characteristic distribution of similar bridge groups, improving the reliability of the reference distribution.

[0081] Step S414: For each embedding cluster, introduce a structural scale adjustment term and a temperature coefficient to adjust the corresponding covariance matrix;

[0082] Understandably, embedded spatial clustering enables the grouping of similar bridges, ensuring the relevance of the reference feature distribution. However, different bridges have different tolerance ranges for deviation. For example, long-span bridges are more sensitive to small deviations, requiring amplified constraints. Thermal expansion and contraction caused by temperature changes can affect joint width and deviation, necessitating the integration of probabilistic modeling. Therefore, a structural scale adjustment term is designed to adapt to bridges of different spans and sizes, and the influence of environmental factors is corrected through a temperature coefficient, making the reference distribution more closely match the actual engineering scenario.

[0083] Specifically, by adjusting the covariance, the direct influence of bridge scale and ambient temperature is eliminated, making the reference distribution more accurate. For each embedding cluster, the covariance matrix is ​​adjusted as follows:

[0084] ;

[0085] In the formula, Indicates the first The covariance matrix adjusted by embedding clustering Indicates the first The covariance matrix of each embedded cluster. This represents the structural scale adjustment term. This represents the temperature coefficient.

[0086] The temperature coefficient is used for linear temperature compensation.

[0087] ;

[0088] In the formula, Indicates the coefficient of thermal expansion sensitivity. This indicates the ambient temperature at the time of closure. Indicates the reference temperature.

[0089] Step S415: Fit the reference feature distribution to each embedding cluster based on the adjusted covariance matrix to form a reference feature distribution space.

[0090] In this step, Gaussian distribution fitting is performed on the embeddings of each embedding cluster:

[0091] ;

[0092] In the formula, Indicates the first Embedding clusters are related to the embedding representation. Reference feature distribution, Indicates a Gaussian distribution. Indicates the first Cluster centers of an embedded cluster, Indicates the first The covariance matrix after embedding and clustering adjustment.

[0093] A reference feature distribution space is constructed by the reference feature distribution of each embedded cluster, which serves as a reference standard for the closure test.

[0094] Step S42: Map the closure feature vector of the bridge to be tested to the embedding space to obtain the embedding representation of the bridge to be tested, and match the nearest embedding cluster in the embedding space;

[0095] In this step, the closure feature vector of the bridge to be tested is mapped using an embedding mapping function. Simultaneously, the nearest embedding cluster in the embedding space is matched.

[0096] ;

[0097] In the formula, The clustering index representing the most recent embedded cluster. This represents the function that takes the minimum value. , This represents the embedded representation of the bridge to be tested. Indicates the first Cluster centers of an embedded cluster, This represents the Euclidean norm.

[0098] Step S43: Determine the corresponding reference feature distribution in the reference feature distribution space through the matched embedding clusters;

[0099] Step S44: Calculate the degree of distribution deviation of the bridge under test by using the embedded representation of the bridge under test and the determined reference feature distribution;

[0100] In this step, the Mahalanobis distance is used to measure the degree of deviation of the bridge under inspection from the reference distribution space. The larger the distance, the more serious the deviation, which may indicate an anomaly in the closure.

[0101] The formula for calculating the degree of distribution deviation is:

[0102] ;

[0103] In the formula, Indicates the degree of distribution deviation. This represents the embedded representation of the bridge to be tested. Indicates the first Cluster centers of an embedded cluster, Indicates the first The covariance matrix adjusted by embedding clustering express The transpose of .

[0104] Step S45: Obtain the closure inspection result of the bridge under inspection by measuring the degree of distribution deviation.

[0105] In this step, the degree of distribution deviation is converted into a qualified or unqualified closure inspection result by using a preset threshold (which can be determined by historical bridge statistics).

[0106] Example 2:

[0107] like Figure 2 As shown in the figure, this embodiment provides an image-based bridge rotation and closure inspection system, the system comprising:

[0108] The acquisition module is used to acquire the final state image of the bridge to be inspected before closure after rotation. The final state image before closure includes a top final state image and a side final state image.

[0109] An extraction and construction module is used to extract the structural geometric features of the closure region based on the pre-closure final state image, and to construct a physical constraint reference system for the closure region through the structural geometric features.

[0110] The mapping module is used to perform feature mapping on the pre-closure final state image under the physical constraint reference frame to generate the closure feature vector of the bridge to be inspected.

[0111] The verification module is used to map the closure feature vector to the reference feature distribution space, calculate the degree of distribution deviation between the vector and the reference feature distribution, and obtain the closure verification result.

[0112] The closure area includes a top closure area and a side closure area, and the extraction construction module includes:

[0113] The positioning unit is used to locate the closure area of ​​the bridge rotation based on the pre-closure final state image;

[0114] The first extraction unit is used to extract top physical quantities in the top closure area, including the top projection joint width curve, the planar edge point set of the top closure joint, and the plane centerline of the main beam axis.

[0115] The second extraction unit is used to extract lateral physical quantities within the lateral closure area. The lateral physical quantities include the vertical actual joint width curve and the boundary elevation point set of the lateral closure joint.

[0116] The first calculation unit is used to calculate the top geometric statistical characteristics through top physical quantities. The top geometric statistical characteristics include the lateral offset of the closure joint centerline in the plane, the plane deflection angle of the main beam axis, and the plane projection deviation index.

[0117] The second calculation unit is used to calculate lateral geometric statistical characteristics through lateral physical quantities. The lateral geometric statistical characteristics include the actual top seam width, the actual bottom seam width, the seam width variation gradient, the vertical misalignment, and the vertical non-uniformity index.

[0118] The building blocks are used to construct structural geometric features using top and lateral geometric statistical features.

[0119] The mapping module includes:

[0120] The first mapping unit is used to map the pixel coordinates in the top final image and the side final image to a unified physical constraint reference system coordinate system to obtain the feature point set of the closure region.

[0121] The projection unit is used to project the geometric features of the structure based on the feature point set of the closure region to obtain the three-dimensional structural quantity;

[0122] The encoding unit is used to combine the three-dimensional structural quantities with the bridge type, span grade, bridge size and ambient temperature at the time of closure of the bridge under inspection to generate the closure feature vector of the bridge under inspection.

[0123] The inspection module includes:

[0124] The second mapping unit is used to map the closure feature vectors of the verified bridge to the embedding space, and to construct a reference feature distribution space based on the embedding space;

[0125] The third mapping unit is used to map the closure feature vector of the bridge to be tested to the embedding space to obtain the embedding representation of the bridge to be tested, and to match the nearest embedding cluster in the embedding space.

[0126] The determining unit is used to determine the corresponding reference feature distribution in the reference feature distribution space through the matched embedding clusters;

[0127] The third calculation unit is used to calculate the degree of distribution deviation of the bridge under test by using the embedded representation of the bridge under test and the determined reference feature distribution.

[0128] The inspection unit is used to obtain the closure inspection results of the bridge under inspection by measuring the degree of distribution deviation.

[0129] It should be noted that the specific methods by which each module performs operations in the system described in the above embodiments have been described in detail in the embodiments related to the method, and will not be elaborated here.

[0130] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

[0131] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An image-based method for inspecting the rotation and closure of bridges, characterized in that, include: Obtain the final state image of the bridge before closure after rotation, including the top final state image and the side final state image; Based on the pre-closure final state image, the structural geometric features of the closure region are extracted, and a physical constraint reference system for the closure region is constructed using the structural geometric features. Under the physical constraint reference frame, feature mapping is performed on the final state image before closure to generate the closure feature vector of the bridge to be tested; The closure feature vector is mapped to a reference feature distribution space, and the degree of distribution deviation between the vector and the reference feature distribution is calculated to obtain the closure test result. The closure region includes a top closure region and a side closure region. Extracting the structural geometric features of the closure region based on the pre-closure final state image includes: Based on the pre-closure final state image, the closure area of ​​the bridge rotation is located; Extract the top physical quantities within the top closure area. The top physical quantities include the top projected joint width curve, the planar edge point set of the top closure joint, and the plane centerline of the main beam axis. Lateral physical quantities are extracted within the lateral closure area, including the actual vertical joint width curve and the set of boundary elevation points of the lateral closure joint; The top geometric statistical characteristics are calculated by measuring the top physical quantities. These top geometric statistical characteristics include the lateral offset of the closure joint centerline in the plane, the plane deflection angle of the main beam axis, and the plane projection deviation index. Lateral geometric statistical characteristics are calculated using lateral physical quantities. These lateral geometric statistical characteristics include the actual top seam width, the actual bottom seam width, the seam width variation gradient, the vertical misalignment, and the vertical non-uniformity index. Structural geometric features are constructed by using top and lateral geometric statistical features; Specifically, the closure feature vector is mapped to a reference feature distribution space, and the degree of distribution deviation between the vector and the reference feature distribution is calculated to obtain the closure verification result, including: The closure feature vectors of the verified bridges are mapped to the embedding space, and a reference feature distribution space is constructed based on the embedding space. The closure feature vector of the bridge to be tested is mapped to the embedding space to obtain the embedding representation of the bridge to be tested, and the nearest embedding cluster in the embedding space is matched. The corresponding reference feature distribution in the reference feature distribution space is determined by the matched embedding clusters; The degree of distribution deviation of the bridge under test is calculated by using the embedded representation of the bridge under test and the determined reference feature distribution. The closure test results of the bridge under test are obtained by measuring the degree of distribution deviation. The process of mapping the closure feature vectors of the verified bridge to an embedding space and constructing a reference feature distribution space based on the embedding space includes: Obtain the closure feature vector of the verified bridge; The closure feature vectors of the tested bridges are mapped to the embedding space to obtain the embedding representation of each tested bridge; Clustering the embedded representations of the verified bridges yields multiple embedding clusters; For each embedded cluster, a structural scale adjustment term and a temperature coefficient are introduced to adjust the corresponding covariance matrix; Based on the adjusted covariance matrix, a reference feature distribution is fitted to each embedding cluster to form a reference feature distribution space; The formula for calculating the degree of distribution deviation is: ； In the formula, Indicates the degree of distribution deviation. This represents the embedded representation of the bridge to be tested. Indicates the first Cluster centers of an embedded cluster, Indicates the first The covariance matrix adjusted by embedding clustering express The transpose of .

2. The image-based bridge rotation and closure inspection method according to claim 1, characterized in that, The construction of the physical constraint reference system for the closure region through the structural geometric features includes: Based on the centerline of the main beam axis plane, the main structure reference axis is determined, and the main structure reference axis serves as the longitudinal reference axis of the physical constraint reference system; Based on the main structure reference axis, a planar constraint reference axis orthogonal to the main structure reference axis is determined in the bridge deck plane defined by the top closure area; The vertical physical constraint axis is determined by the set of boundary elevation points of the vertical misalignment and the lateral closure joint. Using the main structure reference axis, the planar constraint reference axis, and the vertical physical constraint axis as coordinate references, a physical constraint reference system for the closure region is constructed.

3. The image-based bridge rotation and closure inspection method according to claim 1, characterized in that, The step of performing feature mapping on the pre-closure final state image under the physical constraint reference frame to generate the closure feature vector of the bridge to be inspected includes: The pixel coordinates in the top final image and the side final image are mapped to a unified physical constraint reference system coordinate system to obtain the feature point set of the closure region; The geometric features of the structure are projected based on the feature point set of the closure region to obtain the three-dimensional structural quantity. After combining the three-dimensional structural quantities with the bridge type, span grade, bridge size, and ambient temperature at the time of closure of the bridge under test, a closure feature vector of the bridge under test is generated.

4. An image-based bridge rotation and closure inspection system, characterized in that, include: The acquisition module is used to acquire the final state image of the bridge to be inspected before closure after rotation. The final state image before closure includes a top final state image and a side final state image. An extraction and construction module is used to extract the structural geometric features of the closure region based on the pre-closure final state image, and to construct a physical constraint reference system for the closure region through the structural geometric features. The mapping module is used to perform feature mapping on the pre-closure final state image under the physical constraint reference frame to generate the closure feature vector of the bridge to be inspected. The verification module is used to map the closure feature vector to the reference feature distribution space, calculate the degree of distribution deviation between the vector and the reference feature distribution, and obtain the closure verification result. The closure area includes a top closure area and a side closure area, and the extraction construction module includes: The positioning unit is used to locate the closure area of ​​the bridge rotation based on the pre-closure final state image; The first extraction unit is used to extract top physical quantities in the top closure area, including the top projection joint width curve, the planar edge point set of the top closure joint, and the plane centerline of the main beam axis. The second extraction unit is used to extract lateral physical quantities within the lateral closure area. The lateral physical quantities include the vertical actual joint width curve and the boundary elevation point set of the lateral closure joint. The first calculation unit is used to calculate the top geometric statistical characteristics through top physical quantities. The top geometric statistical characteristics include the lateral offset of the closure joint centerline in the plane, the plane deflection angle of the main beam axis, and the plane projection deviation index. The second calculation unit is used to calculate lateral geometric statistical characteristics through lateral physical quantities. The lateral geometric statistical characteristics include the actual top seam width, the actual bottom seam width, the seam width variation gradient, the vertical misalignment, and the vertical non-uniformity index. Building units are used to construct structural geometric features using top and lateral geometric statistical features; The inspection module includes: The second mapping unit is used to map the closure feature vectors of the verified bridge to the embedding space, and to construct a reference feature distribution space based on the embedding space; The third mapping unit is used to map the closure feature vector of the bridge to be tested to the embedding space to obtain the embedding representation of the bridge to be tested, and to match the nearest embedding cluster in the embedding space. The determining unit is used to determine the corresponding reference feature distribution in the reference feature distribution space through the matched embedding clusters; The third calculation unit is used to calculate the degree of distribution deviation of the bridge under test by using the embedded representation of the bridge under test and the determined reference feature distribution. The inspection unit is used to obtain the closure inspection results of the bridge under inspection by measuring the degree of distribution deviation; The process of mapping the closure feature vectors of the verified bridge to an embedding space and constructing a reference feature distribution space based on the embedding space includes: Obtain the closure feature vector of the verified bridge; The closure feature vectors of the tested bridges are mapped to the embedding space to obtain the embedding representation of each tested bridge; Clustering the embedded representations of the verified bridges yields multiple embedding clusters; For each embedded cluster, a structural scale adjustment term and a temperature coefficient are introduced to adjust the corresponding covariance matrix; Based on the adjusted covariance matrix, a reference feature distribution is fitted to each embedding cluster to form a reference feature distribution space; The formula for calculating the degree of distribution deviation is: ； In the formula, Indicates the degree of distribution deviation. This represents the embedded representation of the bridge to be tested. Indicates the first Cluster centers of an embedded cluster, Indicates the first The covariance matrix adjusted by embedding clustering express The transpose of .

5. The image-based bridge rotation and closure inspection system according to claim 4, characterized in that, The mapping module includes: The first mapping unit is used to map the pixel coordinates in the top final image and the side final image to a unified physical constraint reference system coordinate system to obtain the feature point set of the closure region. The projection unit is used to project the geometric features of the structure based on the feature point set of the closure region to obtain the three-dimensional structural quantity; The encoding unit is used to combine the three-dimensional structural quantities with the bridge type, span grade, bridge size and ambient temperature at the time of closure of the bridge under inspection to generate the closure feature vector of the bridge under inspection.

Images