High-speed railway line protection section monitoring method, device, equipment and product

Abstract

The application provides a high-speed rail line protection section monitoring method, device, equipment and product, comprising: acquiring multi-period point cloud data of a high-speed rail line protection section and line geographic information collected by an airborne laser radar at different monitoring moments, and dividing monitoring units according to line mileage, and determining candidate object regions corresponding to different object categories in each mileage monitoring unit; extracting object feature points of corresponding types from line infrastructure objects in the candidate object regions according to the object categories, and obtaining an object feature point set; screening same-name object feature point pairs between different monitoring moments according to line mileage, object categories, feature point types and local geometric features corresponding to each object feature point; performing object semantic constraint registration on the multi-period point cloud data based on the same-name object feature point pairs; determining local deformation data and overall deformation data of the line infrastructure objects according to coordinate differences of each same-name object feature point in the same line infrastructure object after registration, and determining deformation monitoring results of the high-speed rail line protection section based on the local deformation data and the overall deformation data.

Description

Technical Field

[0001] This invention relates to the field of railway infrastructure monitoring technology, and in particular to a method, device, equipment and product for monitoring protected sections of high-speed railway lines. Background Technology

[0002] As the operational mileage of high-speed railway lines increases, the infrastructure along these lines is susceptible to changes in geological conditions, rainfall erosion, temperature variations, train dynamic loads, and disturbances from the surrounding environment during long-term service. This can lead to varying degrees of settlement, displacement, cracking, or localized deformation. Particularly in monitored sections of high-speed railway lines, such as slopes, bridges, tunnel entrances, track structures, and areas prone to geological disasters, the diverse types of infrastructure, continuous spatial distribution, and complex surrounding environments mean that even minor deformations, if not identified in time, can escalate into hazards affecting line stability and operational safety. Therefore, continuous and accurate deformation monitoring of multiple infrastructure objects within the monitored sections of high-speed railway lines is necessary to provide a data foundation for line condition assessment and maintenance.

[0003] While existing methods such as manual inspection, total station measurement, periodic testing, and conventional 3D laser scanning can obtain some status information, they still have the following shortcomings in the large-scale, multi-object monitoring scenarios of high-speed railway protection sections: First, high-speed railway protection sections typically include various infrastructure objects such as bridges, slopes, and tracks. Different objects have different spatial morphologies and stability characteristics. If the overall point cloud is directly registered, it is easy to establish incorrect correspondences between different objects, leading to the accumulation of registration errors. Second, existing point cloud registration methods usually rely on the spatial proximity relationship between point clouds and do not fully incorporate constraints such as line mileage, object category, and object feature information. In cases of long-distance strip lines, changes in point cloud density, or local occlusion, it is difficult to stably determine the same feature points at different monitoring times, affecting the accuracy of deformation detection. In addition, existing methods usually make status judgments based on differences in the overall point cloud or changes in discrete points, making it difficult to accurately associate the detection results with specific line infrastructure objects and their local structural locations. Summary of the Invention

[0004] This invention provides a method, device, equipment, and product for monitoring protected sections of high-speed railway lines, which solves the problem that existing point cloud monitoring methods do not fully utilize line mileage, object category, and object feature information for constraints in high-speed railway line protected sections, which contain multiple types of line infrastructure objects, leading to mismatches during multi-stage point cloud registration and affecting the accuracy of deformation detection of line infrastructure objects.

[0005] This invention provides a method for monitoring protected sections of high-speed railway lines, comprising: acquiring multi-period point cloud data and line geographic information of the protected section of a high-speed railway line collected by airborne lidar at different monitoring times; determining multiple mileage monitoring units based on the line geographic information according to the line mileage and facility boundaries, and determining candidate object areas corresponding to different object categories within each mileage monitoring unit; extracting object feature points of the type corresponding to the object category from the line infrastructure objects in the candidate object areas according to the object category; filtering pairs of identical object feature points at different monitoring times based on the line mileage, object category, feature point type, and local geometric features corresponding to each object feature point; performing object semantic constraint weighted registration on the multi-period point cloud data according to the object category weight and the matching confidence of each pair of identical object feature points based on the identical object feature points; determining the local deformation data and overall deformation data of the line infrastructure object based on the coordinate differences of the identical object feature points in the same line infrastructure object after registration, and determining the deformation monitoring result of the protected section of the high-speed railway line based on the local deformation data and overall deformation data.

[0006] According to the high-speed railway line protection section monitoring method provided by the present invention, the step of determining multiple mileage monitoring units based on the line geographic information according to the line mileage and facility boundaries, and determining candidate object areas corresponding to different object categories within each of the mileage monitoring units, includes: unifying the point cloud data with the high-speed railway line centerline, mileage information, and design geometric areas of line infrastructure objects into the same spatial coordinate system according to the line geographic information; determining multiple mileage monitoring units according to a preset mileage length or facility boundaries of line infrastructure objects; and determining candidate object areas corresponding to each object category based on the mileage difference between the line mileage corresponding to the point cloud data and the mileage monitoring units, and the spatial distance between the point cloud data and the design geometric areas of the corresponding object categories.

[0007] According to the high-speed railway line protection section monitoring method provided by the present invention, the object category includes at least one of bridge support objects, slope objects, bridge body objects, and track objects; the step of extracting object feature points of the corresponding type from the line infrastructure objects in the candidate object area according to the object category includes: when the object category is a bridge support object, extracting the end feature points and contour feature points distributed along the height direction of the bridge support object; when the object category is a slope object, extracting the slope boundary feature points, elevation extreme value feature points, and grid sampling feature points of the slope object; when the object category is a bridge body object, extracting the joint feature points and contour corner feature points of the bridge body object; and when the object category is a track object, extracting the centerline feature points, joint feature points, and rail surface feature points of the track object.

[0008] According to the high-speed railway line protection section monitoring method provided by the present invention, the step of filtering the same object feature point pairs between different monitoring times based on the line mileage, object category, feature point type and local geometric features corresponding to each object feature point includes: obtaining the line mileage, object category, feature point type, spatial coordinates, normal vector and curvature corresponding to each object feature point; when two object feature points corresponding to different monitoring times have the same object category and the same feature point type, and the corresponding line mileage difference, normal vector difference and curvature difference respectively meet the corresponding preset difference thresholds, the two object feature points are determined as the same object feature point pair.

[0009] According to the high-speed railway line protection section monitoring method provided by the present invention, the step of performing object semantic constraint weighted registration on the point cloud data of multiple periods based on the same object feature point pairs, according to the object category weight and the matching confidence of each same object feature point pair, includes: determining the matching confidence of each same object feature point pair based on the line mileage difference, curvature difference and normal vector consistency corresponding to each same object feature point pair; weighting the registration error of each same object feature point pair based on the object category weight and the matching confidence of each same object feature point pair; determining the rotation matrix and translation vector between the point cloud data at different monitoring times based on the weighted registration error, and performing coordinate transformation on at least one period of the point cloud data according to the rotation matrix and translation vector to obtain the registered multi-period point cloud data.

[0010] According to the high-speed railway line protection section monitoring method provided by the present invention, the step of determining the deformation monitoring result of the high-speed railway line protection section based on the local deformation data and the overall deformation data includes: determining the three-dimensional deformation of each of the corresponding object feature points according to the spatial coordinates of each of the corresponding object feature points at different monitoring times; determining the overall deformation of the line infrastructure object according to the three-dimensional deformation of each of the corresponding object feature points in the same line infrastructure object and the corresponding stability weight, and determining the maximum local deformation of the line infrastructure object according to the three-dimensional deformation of each of the corresponding object feature points; determining the deformation rate according to the overall deformation of the line infrastructure object at continuous monitoring times; determining the risk score of the corresponding line infrastructure object based on the overall deformation, the maximum local deformation and the deformation rate, and determining the risk level of the line infrastructure object according to the risk score.

[0011] The high-speed railway line protection section monitoring method provided by the present invention further includes: performing point cloud filtering and noise removal on the point cloud data of each period, and performing point cloud classification and segmentation and geometric feature extraction on the processed point cloud data to obtain preprocessed point cloud data of each period; and determining the candidate object region based on the corresponding preprocessed point cloud data.

[0012] This invention also provides a high-speed railway line protection section monitoring device, comprising: an information acquisition module for acquiring multi-phase point cloud data and line geographic information of the high-speed railway line protection section collected by airborne lidar at different monitoring times; a classification module for determining multiple mileage monitoring units according to the line geographic information, based on the line mileage and facility boundaries, and determining candidate object areas corresponding to different object categories within each mileage monitoring unit; a feature point extraction module for extracting object feature points of the type corresponding to the object category from the line infrastructure objects in the candidate object areas according to the object category; and a filtering module for filtering according to the object features. The system uses a point cloud data collection module to filter pairs of identical object feature points across different monitoring times, based on the corresponding line mileage, object category, feature point type, and local geometric features. A registration module performs object semantic constraint weighted registration on the point cloud data from multiple periods, according to the object category weight and the matching confidence of each identical object feature point pair. A monitoring module determines the local deformation data and overall deformation data of the line infrastructure object based on the coordinate differences of the identical object feature points within the same registered line infrastructure object, and determines the deformation monitoring results of the high-speed railway line protection section based on the local deformation data and overall deformation data.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the high-speed railway line protection section monitoring method described above.

[0014] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the high-speed railway line protection section monitoring method as described above.

[0015] The high-speed railway line protection section monitoring method, device, equipment, and product provided by this invention distinguishes railway infrastructure objects in multi-phase point cloud data by line mileage and object category, and extracts corresponding object feature points for different object categories. Furthermore, it combines line mileage, object category, feature point type, and local geometric features to screen feature point pairs of objects with the same name, and performs semantically constrained weighted registration accordingly. This reduces erroneous matching between different railway infrastructure objects and improves the stability and accuracy of multi-phase point cloud registration. Based on this, it determines local deformation data and overall deformation data according to the coordinate differences of feature points of objects with the same name within the same railway infrastructure object, enabling deformation results to correspond to specific objects and structural locations, thereby improving the accuracy and reliability of deformation monitoring results for high-speed railway line protection sections. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0017] Figure 1 This is a flowchart of the high-speed railway line protection section monitoring method of the present invention; Figure 2 The flowchart illustrates an example of the present invention, which involves determining multiple mileage monitoring units based on route geographic information, according to route mileage and facility boundaries, and determining candidate object areas corresponding to different object categories within each mileage monitoring unit. Figure 3 The flowchart illustrates an example of the present invention of performing object semantic constraint weighted registration on multi-period point cloud data based on feature point pairs of objects with the same name, according to the object category weight and the matching confidence of each feature point pair of objects with the same name; Figure 4 A flowchart illustrating an example of the present invention is provided, showing how to determine the deformation monitoring results of a high-speed railway line protection section based on local deformation data and overall deformation data. Figure 5A structural block diagram of the high-speed railway line protection section monitoring device of the present invention is shown; Figure 6 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this 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 this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0019] Figure 1 This is a flowchart of the high-speed railway line protection section monitoring method of the present invention.

[0020] like Figure 1 As shown, the method includes operations S110~S160.

[0021] The S110 system acquires multi-phase point cloud data and line geographic information of the high-speed railway protection section collected by airborne lidar at different monitoring times.

[0022] According to an embodiment of the present invention, the protected section of a high-speed railway line can be a section along the high-speed railway line that requires continuous deformation monitoring, which may contain bridges, slopes, tracks, roadbeds, tunnel entrances, and other line infrastructure objects that need to be monitored. Airborne lidar can be mounted on a drone or other flight platform to scan along the protected section of the high-speed railway line and acquire point cloud data that characterizes the three-dimensional spatial morphology of the line infrastructure objects.

[0023] Line geographic information is used to characterize the spatial distribution of high-speed railway lines and their associated facilities. It can include information such as the line centerline, line mileage, the location of line infrastructure objects, and the boundaries of corresponding facilities. By establishing a spatial correspondence between multi-period point cloud data and line geographic information, the spatial location in the point cloud can be further associated with specific line mileage and line infrastructure objects.

[0024] According to embodiments of the present invention, multi-period point cloud data refers to point cloud data collected at different monitoring times. These different monitoring times can be different dates, different inspection cycles, or monitoring points following different operating conditions. By acquiring multi-period point cloud data, a data foundation can be provided for subsequent comparisons of the spatial state changes of infrastructure objects in the same area or along the same route at different times.

[0025] In operation S120, based on the geographical information of the line, multiple mileage monitoring units are determined according to the line mileage and facility boundaries, and candidate object areas corresponding to different object categories are determined within each mileage monitoring unit.

[0026] According to embodiments of the present invention, high-speed railway protection zones are typically distributed in a long strip along the railway line, and various railway infrastructure objects may coexist at different locations. If the point cloud data corresponding to the entire protection zone is directly used as a unified processing object, different mileage locations and different types of railway infrastructure objects are easily mixed up. Therefore, the high-speed railway protection zones can be spatially organized first using the railway line's geographical information.

[0027] Specifically, the protection sections of high-speed railway lines can be segmented according to line mileage, and each mileage monitoring unit can be determined in conjunction with the facility boundaries of the line infrastructure objects. The mileage monitoring unit is used to define the line range corresponding to the subsequent object analysis. One mileage monitoring unit can correspond to a line section of predetermined length, or it can be determined based on the spatial distribution range of a certain line infrastructure object or multiple adjacent line infrastructure objects.

[0028] In operation S130, based on the object category, object feature points of the type corresponding to the object category are extracted from the line infrastructure objects in the candidate object region.

[0029] According to embodiments of the present invention, different categories of railway infrastructure objects have different geometric structures and spatial morphologies, and their geometric positions, which stably represent the object's location and structural state in multi-period point cloud data, also differ. Therefore, after determining candidate object regions, feature points can be extracted from the railway infrastructure objects within the candidate object regions using a feature extraction method adapted to that object category.

[0030] Object feature points can be representative points that can characterize the spatial location, outline shape, or local structural features of infrastructure objects. For different object categories, feature points that can relatively stably reflect the geometric shape of the corresponding object can be selected to form object feature points, without having to use the entire point cloud within the candidate object region as the basis for subsequent registration.

[0031] In operation S140, based on the line mileage, object category, feature point type, and local geometric features corresponding to each object feature point, pairs of object feature points with the same name at different monitoring times are selected.

[0032] According to an embodiment of the present invention, a pair of object feature points with the same name refers to two object feature points located in point cloud data at different monitoring times, corresponding to the same or corresponding structural positions within the same railway infrastructure object. Since there may be a large number of railway infrastructure objects with similar geometric shapes within the protection zone of a high-speed railway line, determining the correspondence solely based on the spatial distance between object feature points may lead to incorrect matching between adjacent mileage locations or different facility objects. Therefore, in this embodiment, it is necessary to combine multiple attribute information corresponding to the object feature points to filter candidate correspondences.

[0033] Specifically, for object feature points extracted at different monitoring times, the corresponding line location can first be defined based on the line mileage, and then it can be determined whether two object feature points have the same object semantics and structural meaning based on the object category and feature point type. Furthermore, by combining the local geometric features reflected in the neighborhood of the object feature points, object feature points with similar geometric states can be matched. Candidate matching relationships that are significantly inconsistent in line location, object semantics, feature point type, or local geometric state can be excluded.

[0034] Therefore, feature point pairs with high correspondence reliability can be obtained from the feature points of objects at different monitoring times. This makes subsequent point cloud registration no longer solely dependent on the spatial proximity between points, but also constrained by both the line location and the semantics of the objects, thereby reducing the possibility of incorrect matching across mileage, across objects, and across structural locations.

[0035] In operation S150, based on feature point pairs of objects with the same name, and according to the object category weight and the matching confidence of each feature point pair of objects with the same name, the multi-period point cloud data is subjected to object semantic constraint weighted registration.

[0036] According to an embodiment of the present invention, point cloud data corresponding to one monitoring time can be selected as reference data, and point cloud data corresponding to other monitoring times can be selected as data to be registered. Based on the selected pairs of feature points of the same object, the spatial transformation relationship between the data to be registered and the reference data is determined, so as to transform the point cloud data from multiple periods to a unified spatial coordinate reference.

[0037] Considering the differences in geometric stability and feature saliency among different categories of line infrastructure objects, corresponding object category weights can be assigned to different object categories. Furthermore, since the matching reliability of feature point pairs of different objects with the same name varies, the weights of each feature point pair in the registration process can be adjusted based on the corresponding matching confidence level.

[0038] Furthermore, the registration errors of feature point pairs of each identical object are weighted according to the object category weight and matching confidence, and the spatial transformation relationship is determined based on the weighted overall registration error. The coordinates of the corresponding point cloud data are then transformed according to the spatial transformation relationship to obtain registered multi-period point cloud data.

[0039] In operation S160, based on the coordinate differences of feature points of each object with the same name in the same line infrastructure object after registration, the local deformation data and overall deformation data of the line infrastructure object are determined, and the deformation monitoring results of the high-speed railway line protection section are determined based on the local deformation data and overall deformation data.

[0040] According to an embodiment of the present invention, after completing the registration of multi-period point cloud data, the spatial coordinates of the same object feature points at different monitoring times can be compared, and the displacement changes of each object feature point can be determined based on the coordinate differences.

[0041] According to embodiments of the present invention, local deformation data of a railway infrastructure object is determined based on the displacement changes of a single object feature point or multiple object feature points within a local area; and overall deformation data is determined by combining the displacement changes of multiple object feature points within the railway infrastructure object. Thus, the deformation of both the local location and the overall structure of the railway infrastructure object can be reflected respectively.

[0042] Based on the local and overall deformation data of each line infrastructure object, the deformation state of the corresponding line location can be determined, and deformation monitoring results of the high-speed railway line protection section can be generated.

[0043] In one illustrative embodiment, point cloud filtering and noise removal can be performed on the point cloud data of each period, and point cloud classification, segmentation and geometric feature extraction can be performed on the processed point cloud data to obtain preprocessed point cloud data of each period, and candidate object regions can be determined based on the corresponding preprocessed point cloud data.

[0044] According to an embodiment of the present invention, after acquiring point cloud data for each period, the point cloud data for each period can be preprocessed first to reduce the noise generated during the acquisition process and the impact of non-line infrastructure object point clouds on subsequent processing.

[0045] Specifically, ground point filtering can be performed on the point cloud data of each period to separate effective point cloud data related to the terrain surface and line infrastructure objects from the original point cloud data.

[0046] In one example, a progressive triangulation filtering method can be used for ground point filtering. Specifically, initial ground seed points can be selected from the point cloud data first, and an initial irregular triangulation can be constructed based on the initial ground seed points; then, for the point to be judged, its vertical distance to the corresponding triangular face can be calculated, and the vertical distance can be used to determine whether the point to be judged meets the ground point judgment condition.

[0047] For example, let the three vertices of the triangle be A, B, and C, and let P be the point to be determined. The normal vector of the triangle is... Then the perpendicular distance d from point P to the triangle face can be expressed as:

[0048] Furthermore, the average distance distribution can be determined based on the average distance of all point cloud points, and the distance judgment threshold can be determined based on the mean and standard deviation of the average distance. If the average distance corresponding to a certain point exceeds the distance judgment threshold, it can be considered that the spatial continuity between that point and the surrounding point cloud is poor, and it can be identified as an outlier and removed.

[0049] In another example, a neighborhood noise removal method can be used to filter out locally sparse points. Specifically, a search radius and a minimum number of neighborhood points can be set. For any point in the point cloud data, the number of neighborhood points within its search radius is counted. If the number of neighborhood points is less than a preset threshold, the point can be identified as a noise point and removed.

[0050] Through the above filtering and noise removal processes, the impact of terrain undulations, isolated points, and local sparse noise on the subsequent point cloud classification, segmentation, and object feature extraction processes can be reduced, resulting in denoised point cloud data.

[0051] Furthermore, the denoised point cloud data can be classified and segmented to obtain point cloud regions corresponding to different spatial structures.

[0052] Specifically, based on the spatial distance, normal vector difference, curvature difference, and spatial continuity between adjacent points in the point cloud data, region growing or clustering segmentation can be performed on the denoised point cloud data to divide point clouds with similar spatial structures and geometric features into the same point cloud region.

[0053] After completing the point cloud classification and segmentation, geometric features can be extracted from each point cloud region to obtain geometric features that characterize the spatial morphology of the line infrastructure object.

[0054] In one example, the covariance matrix can be constructed based on the neighboring points of each point within the point cloud region:

[0055] in, Let C be the mean coordinates of the neighborhood point set. Eigenvalue decomposition of the covariance matrix C yields multiple eigenvalues ​​and their corresponding eigenvectors. The eigenvector corresponding to the smallest eigenvalue can be used as the normal vector of that point, and the eigenvector corresponding to the largest eigenvalue as the principal direction of that point. The curvature of the point can then be determined based on these eigenvalues. For example, let the eigenvalues ​​be... ,and Then curvature It can be represented as:

[0056] By classifying and segmenting the denoised point cloud data and extracting geometric features, we can obtain preprocessed point cloud data for each period that contains spatial distribution information and geometric structure information.

[0057] Furthermore, candidate object regions can be determined based on the corresponding preprocessed point cloud data.

[0058] Specifically, preprocessed point cloud data can be spatially correlated with route geographic information, and the preprocessed point cloud data can be regionalized based on route mileage, spatial location of route infrastructure objects, and facility boundary information.

[0059] For example, the corresponding route range can be determined based on the route centerline and mileage information, and candidate object regions for the corresponding object categories can be determined in the preprocessed point cloud data by combining the spatial distribution characteristics of route infrastructure objects such as bridges, slopes, and tracks.

[0060] Figure 2 The flowchart illustrates an example of the present invention, which involves determining multiple mileage monitoring units based on route geographic information, according to route mileage and facility boundaries, and determining candidate object areas corresponding to different object categories within each mileage monitoring unit.

[0061] like Figure 2 As shown, operation S120 includes operations S210~S230.

[0062] When operating S210, point cloud data is unified with the high-speed rail line centerline, mileage information, and the design geometric area of ​​the line infrastructure objects into the same spatial coordinate system based on the line geographic information.

[0063] According to embodiments of the present invention, the route geographic information may include the high-speed railway line centerline, route mileage information, bridge locations, slope ranges, track design locations, and other design geometric areas corresponding to route infrastructure objects. By performing spatial coordinate transformation between point cloud data and route geographic information, a correspondence can be established between the spatial locations in the point cloud data and the route mileage locations and the design areas of the route infrastructure objects.

[0064] When operating S220, multiple mileage monitoring units are determined according to the preset mileage length or the facility boundary of the line infrastructure object.

[0065] According to an embodiment of the present invention, in order to avoid processing the point cloud data of long-distance high-speed railway lines as a whole, the protection sections of the high-speed railway lines can be divided according to the line mileage information to obtain multiple mileage monitoring units.

[0066] Specifically, the high-speed rail line protection zone can be divided into sections along the centerline of the line according to a preset mileage length. For example, a continuous line can be divided into multiple mileage intervals according to a fixed length. Alternatively, it can be divided according to the facility boundaries of the line infrastructure objects, so that each mileage monitoring unit corresponds to one or more line infrastructure object areas with a clear spatial range.

[0067] For example, for bridge sections, the corresponding mileage monitoring units can be determined based on the facility boundaries corresponding to the bridge body and bridge supports; for slope sections, the corresponding mileage monitoring units can be determined based on the slope protection range; for continuous track sections, the corresponding mileage monitoring units can be divided according to the preset mileage length.

[0068] In operation S230, the candidate object region corresponding to each object category is determined based on the mileage difference between the line mileage corresponding to the point cloud data and the mileage monitoring unit, as well as the spatial distance between the point cloud data and the design geometric region of the corresponding object category.

[0069] According to an embodiment of the present invention, after determining multiple mileage monitoring units, the line mileage constraints and object space constraints can be further combined to filter candidate object regions corresponding to different line infrastructure objects from point cloud data.

[0070] Let any point in the cloud be... The corresponding route mileage is The object category is c Object category c The corresponding design geometry region is The center of the mileage segment is s Mileage tolerance is The spatial buffer distance is Then the object category c In mileage section s Candidate point cloud region within Represented as:

[0071] in, Represents point cloud data points With object category c Corresponding design geometric area The spatial distance between them.

[0072] When a point cloud data point simultaneously satisfies the following conditions: the mileage difference between its corresponding line mileage and the current mileage monitoring unit is less than or equal to the preset mileage tolerance, and the spatial distance between the point cloud data point and the corresponding object category's designed geometric area is less than or equal to the preset spatial buffer distance, then the point cloud data point is classified into the candidate object area of ​​the corresponding object category.

[0073] By using the above method, point cloud data can be subject to dual spatial constraints using route mileage information and the design area of ​​route infrastructure objects. This allows point cloud data corresponding to different types of route infrastructure objects, such as bridges, slopes, and tracks, to be pre-divided into corresponding candidate areas in a large-scale route scene, avoiding confusion between point clouds of different objects during subsequent object recognition and deformation analysis.

[0074] In one illustrative embodiment, the object category includes at least one of bridge support objects, slope objects, bridge body objects, and track objects.

[0075] The process of extracting object feature points of the corresponding type from line infrastructure objects in the candidate object region based on the object category may specifically include: When the object category is bridge support, extract the end feature points and contour feature points distributed along the height direction of the bridge support object.

[0076] According to embodiments of the present invention, bridge support objects may include piers, abutments, support columns, or other vertical load-bearing structures used to support the main body of a bridge.

[0077] According to embodiments of the present invention, end feature points may include representative points of the top region and representative points of the bottom region of a bridge support object. For example, the top and bottom point cloud regions of the bridge support object can be determined based on point cloud elevation values, and elevation extreme points, region center points, or contour boundary points can be selected as end feature points in the top and bottom point cloud regions, respectively. Contour feature points distributed along the height direction can be feature points extracted layer by layer along the height direction on the outer contour of the bridge support object. For example, the bridge support object can be divided into multiple height layers according to the elevation direction, and outer contour points, contour center points, or contour extreme points can be extracted in each height layer to form contour feature points continuously distributed along the height direction. Through the above method, the vertical offset, tilt deformation, or local contour changes of the bridge support object can be identified in subsequent registration and deformation calculation processes.

[0078] When the object category is slope, extract the slope boundary feature points, elevation extreme value feature points, and grid sampling feature points of the slope object.

[0079] According to embodiments of the present invention, slope-like objects are typically distributed in a continuous planar pattern, and their deformation may manifest as slope settlement, local bulging, slope toe displacement, or boundary line changes.

[0080] According to embodiments of the present invention, slope boundary feature points may include the boundary points between the slope surface and the roadbed, ground surface, protective structure, or other adjacent structures. For example, the slope boundary line can be determined based on abrupt changes in the normal vector, curvature, or spatial height of the slope point cloud, and multiple boundary sampling points can be extracted from the slope boundary line as slope boundary feature points. Elevation extreme value feature points may include the maximum elevation point, minimum elevation point, or local elevation extreme point in the slope object, used to characterize the overall undulation and local anomaly changes of the slope. Grid sampling feature points may be representative points obtained by sampling within the point cloud area of ​​the slope object according to a preset grid.

[0081] When the object category is a bridge main body object, extract the joint feature points and contour corner feature points of the bridge main body object.

[0082] According to embodiments of the present invention, the main body object of a bridge may include the beam, the bridge deck structure, the bridge joint area, or the bridge edge structure. This type of object typically has relatively obvious structural boundaries, joint locations, and contour corner locations. These locations are relatively easy to form stable correspondences in multi-period point cloud data, and therefore can serve as a source of object feature points.

[0083] According to embodiments of the present invention, joint feature points can be determined based on abrupt changes in normal vectors, curvature, point cloud height, or local gap features in the bridge main body object. For example, regions with curvature greater than a preset curvature condition can be searched in the point cloud of the bridge main body object, or regions with geometric discontinuities along the bridge's extension direction can be searched, and representative points within these regions can be used as joint feature points. Contour corner feature points can be determined based on the outer contour line, edge line, or geometric extreme points of the bridge main body object. For example, contour extreme points in the horizontal, vertical, or longitudinal directions of the bridge main body object can be extracted, or contour line intersections, edge transitions, or corner points can be used as contour corner feature points. Through these methods, the relatively stable joint and corner positions in the bridge main structure can be utilized to provide a basis for multi-stage point cloud registration and bridge main body deformation calculation.

[0084] When the object type is a track object, extract the centerline feature points, joint feature points, and track surface feature points of the track object.

[0085] According to embodiments of the present invention, track-type objects typically extend continuously along the direction of high-speed rail lines, and their deformation monitoring requires attention to changes in the track center position, rail surface elevation, and local changes at joints or connections. Therefore, centerline feature points, joint feature points, and rail surface feature points can be extracted from track-type objects.

[0086] According to embodiments of the present invention, centerline feature points can be determined based on the cross-sectional morphology of the point cloud of the track-like object or the extension characteristics along the line direction. For example, the point cloud of the track-like object can be segmented along the line mileage direction, and centerline feature points can be determined within each segment based on the lateral distribution center, rail surface profile center, or fitting center position of the track point cloud. Joint feature points can be determined based on abrupt curvature changes, local gaps, connection structure positions, or geometric discontinuities in the point cloud within the track-like object. Rail surface feature points can be extracted from the point cloud of the upper surface of the track-like object. For example, rail surface points can be filtered based on elevation, normal vector direction, or rail surface flatness, and multiple rail surface feature points can be sampled along the line direction or lateral position. Through the above processing, the planar offset, vertical settlement, and local joint deformation of the track-like object can be identified in the subsequent deformation calculation process.

[0087] In one illustrative embodiment, operation S140 may specifically include: Obtain the line mileage, object category, feature point type, spatial coordinates, normal vector, and curvature corresponding to each object's feature points.

[0088] If two object feature points at different monitoring times have the same object category and the same feature point type, and the corresponding line mileage difference, normal vector difference, and curvature difference meet the corresponding preset difference thresholds, the two object feature points are identified as a pair of object feature points with the same name.

[0089] According to an embodiment of the present invention, after extracting the object feature points corresponding to each line infrastructure object according to the object category, each object feature point can be further described for matching the corresponding feature positions in the same line infrastructure object at different monitoring times.

[0090] Specifically, for any object feature point, the corresponding feature information such as line mileage, object category, feature point type, spatial coordinates, normal vector, and curvature can be obtained, and an object feature point descriptor can be constructed based on the above feature information.

[0091] For example, suppose the first i The descriptor for each object feature point is:

[0092] in, The feature points correspond to the line mileage. For object categories, For feature point type, ( , ) represents spatial coordinates. It is the normal vector. For curvature.

[0093] After describing the object feature points corresponding to different monitoring times, we can determine the object feature point pairs with the same name based on the consistency relationship between the object feature point descriptors corresponding to different monitoring times.

[0094] Specifically, for two candidate feature points at different monitoring times ( a ) and ( b When the following conditions are met, they are identified as feature point pairs of the same name object: , ,

[0095] ,

[0096] in, , , These are the thresholds for mileage difference, normal vector difference, and curvature difference, respectively. By combining constraints based on object category, feature point type, line mileage, and local geometric features, incorrect matching point pairs across objects and structural locations can be eliminated.

[0097] In other words, based on the premise that two object feature points have the same object category and the same feature point type, it is further determined whether the positional differences, local surface direction differences, and local geometric curvature differences of the two object feature points in the line direction satisfy the corresponding threshold conditions. When all the above conditions are satisfied, it can be considered that the two object feature points correspond to the same structural position or the same representative position in the same line infrastructure object, and the two object feature points are identified as a pair of object feature points with the same name.

[0098] For example, for track-type objects, centerline feature points with the same or similar track mileage at different monitoring times can be used as candidate matching objects, and further filtering can be performed based on differences in normal vectors and curvature; for bridge support-type objects, end feature points on the same pier or contour feature points at the same height can be used as candidate matching objects; for slope-type objects, boundary feature points or grid sampling feature points within the same slope area can be used as candidate matching objects.

[0099] The above method does not directly perform nearest neighbor matching based on the entire point cloud. Instead, it establishes joint constraint relationships using object categories, feature point types, line mileage, and local geometric features. This reduces the possibility of incorrect matching between different line infrastructure objects and between different structural locations, and provides a reliable data foundation for subsequent multi-stage point cloud registration based on feature point pairs of objects with the same name.

[0100] Figure 3 The diagram illustrates a flowchart of an example of the present invention, which describes a weighted registration of multi-period point cloud data based on feature point pairs of objects with the same name, according to the object category weight and the matching confidence of each feature point pair of objects with the same name.

[0101] like Figure 3 As shown, operation S150 includes operations S310 to S330.

[0102] In operation S310, the matching confidence of each pair of feature points of the same object is determined based on the line mileage difference, curvature difference and normal vector consistency of the corresponding feature point pairs.

[0103] According to an embodiment of the present invention, for any pair of feature points with the same name, the line mileage difference, curvature difference, and normal vector consistency information corresponding to the pair of feature points can be obtained. Specifically, the line mileage difference characterizes the positional deviation of the two feature points along the line direction, the curvature difference characterizes the difference in the local geometric shape of the two feature points, and the normal vector consistency characterizes the degree of consistency in the local surface direction where the two feature points are located.

[0104] For example, suppose that the two object feature points in a pair of object feature points with the same name correspond to the line mileage respectively. , The curvatures are respectively , The normal vectors are respectively , Then, the matching confidence of corresponding feature point pairs of the same object can be determined based on the difference in line mileage, curvature, and consistency of normal vectors:

[0105] in, and These are the normalized parameters for the mileage difference and curvature difference, respectively.

[0106] Using the above method, the smaller the difference in line mileage and curvature, and the more consistent the normal vector directions of the two object feature points, the higher the matching confidence of the corresponding object feature point pair; conversely, the contribution of the corresponding object feature point pair to the subsequent registration process is reduced.

[0107] In operation S320, the registration error of each pair of feature points of the same object is weighted according to the object category weight corresponding to each object category and the matching confidence of each pair of feature points of the same object.

[0108] According to an embodiment of the present invention, after obtaining the matching confidence of each pair of feature points of the same object, the registration error of the pair of feature points of the same object under different object categories can be weighted by combining the object category weights corresponding to different line infrastructure objects.

[0109] Specifically, let the feature points in the reference period point cloud be... The feature points in the point cloud to be registered are The set of feature point pairs of objects with the same name is The object category set is Then, a semantically constrained weighted registration objective function can be constructed:

[0110] in, For rotation matrix, It is a translation vector. Weights for different object categories, The matching confidence score for pairs of feature points with the same name. This is the robust error function.

[0111] According to an embodiment of the present invention, object category weight This is used to characterize the importance of different infrastructure objects during the registration process. For example, corresponding weights can be set based on factors such as the structural stability and feature identifiability of bridge support objects, bridge body objects, slope objects, and track objects.

[0112] By incorporating the object category weight and the matching confidence of feature point pairs of the same object into the registration error calculation, the object feature point pairs with higher stability and higher matching reliability can have a greater constraining effect in the registration process, thereby reducing the impact of mismatches between different objects on the registration results.

[0113] In operation S330, the rotation matrix and translation vector between point cloud data at different monitoring times are determined based on the weighted registration error, and the coordinate transformation of at least one period of point cloud data is performed according to the rotation matrix and translation vector to obtain multi-period registered point cloud data.

[0114] According to an embodiment of the present invention, when only two monitoring time points are included, one period of point cloud data can be selected as a reference period, and the coordinate transformation of the other period of point cloud data can be performed according to the determined rotation matrix and translation vector; when multiple monitoring time points are included, one period of point cloud data can be selected as a unified reference period, and the coordinate transformation of the other monitoring time points can be performed respectively.

[0115] Furthermore, after completing the coordinate transformation, the registration error between the point cloud data at different monitoring times after the transformation can be used to determine whether the preset registration conditions are met. When the registration error meets the preset conditions, the registered multi-period point cloud data is obtained; if the preset registration conditions are not met, the feature point pairs of the same object can be re-determined or the rotation matrix and translation vector can be recalculated until the preset registration conditions are met.

[0116] In this way, the present invention no longer uses unconstrained nearest neighbor registration for the overall point cloud, but establishes corresponding relationships within the same object category, the same feature point type, and the same line mileage range. It also combines object category weight and matching confidence to perform weighted constraints, so that the multi-phase point cloud registration process can adapt to the scenario of mixed line infrastructure objects such as bridges, slopes, and tracks in high-speed rail lines, and improve the stability of multi-phase point cloud data spatial alignment.

[0117] Figure 4 The flowchart illustrates an example of the present invention of determining the deformation monitoring results of a high-speed railway line protection section based on local deformation data and overall deformation data.

[0118] like Figure 4 As shown, operation S160 includes operations S410~S440.

[0119] In operation S410, the three-dimensional deformation of each feature point of the same object is determined based on the spatial coordinates of each feature point at different monitoring times.

[0120] According to an embodiment of the present invention, after completing multi-stage point cloud registration, the spatial coordinates of feature points of the same object at different times are compared. Let the same feature point be at the... Period and the The coordinates for each period are as follows: ,

[0121] Then the three-dimensional deformation vector of this feature point for:

[0122] Its horizontal displacement Vertical displacement and three-dimensional composite displacement They are respectively:

[0123]

[0124]

[0125] In operation S420, the overall deformation of the line infrastructure object is determined based on the three-dimensional deformation and corresponding stability weight of the feature points of each object with the same name in the same line infrastructure object, and the maximum local deformation of the line infrastructure object is determined based on the three-dimensional deformation of the feature points of each object with the same name.

[0126] For example, for a certain line infrastructure object Its overall deformation vector It can be obtained by weighted fusion of multiple feature points within the object:

[0127] in, For stability weights.

[0128] Furthermore, the maximum local deformation of the line infrastructure object can be determined based on the three-dimensional deformation corresponding to the feature points of each object with the same name. :

[0129] In this way, the overall deformation is used to reflect the overall structural change trend of the line infrastructure object, and the maximum local deformation is used to reflect the abnormal deformation that may exist in the local area, thereby avoiding the problem that local anomalies cannot be detected by relying solely on the overall change.

[0130] When operating S430, the deformation rate is determined based on the overall deformation of the line infrastructure object during continuous monitoring.

[0131] According to an embodiment of the present invention, after obtaining the overall deformation of the line infrastructure object at multiple continuous monitoring times, the deformation development trend of the line infrastructure object can be further analyzed.

[0132] Specifically, let the object be... During continuous monitoring and The overall deformations are respectively and Then the deformation rate for:

[0133] In operation S440, the risk score of the corresponding line infrastructure object is determined based on the overall deformation, maximum local deformation and deformation rate, and the risk level of the line infrastructure object is determined based on the risk score.

[0134] According to an embodiment of the present invention, risk scoring It can be represented as:

[0135] in, , , These are the overall deformation threshold, the local deformation threshold, and the deformation rate threshold, respectively. , , These are weighting coefficients. Based on risk score. The size of the data can be used to classify the infrastructure objects along the line into four levels: no warning, slight risk, medium risk, and severe risk, according to a preset risk scoring threshold. Furthermore, it can generate object-level deformation monitoring results for key protection sections of the high-speed railway line.

[0136] Through the above methods, the present invention can simultaneously consider factors such as the overall change trend of the line infrastructure objects, local abnormal changes, and change rate. Compared with judging based solely on the difference of a single point cloud, it can more comprehensively reflect the deformation state of infrastructure objects such as bridges, slopes, and tracks within the protection section of the high-speed railway line, and thereby achieve targeted risk level assessment.

[0137] The monitoring device for high-speed railway protection sections provided by the present invention is described below. The monitoring device for high-speed railway protection sections described below can be referred to in correspondence with the monitoring method for high-speed railway protection sections described above.

[0138] Figure 5 A structural block diagram of the high-speed railway line protection section monitoring device of the present invention is shown.

[0139] like Figure 5 As shown, the high-speed railway line protection section monitoring device 500 includes an information acquisition module 510, a classification module 520, a feature point extraction module 530, a screening module 540, a registration module 550, and a monitoring module 560.

[0140] The information acquisition module 510 is used to acquire multi-phase point cloud data and line geographic information of the high-speed railway line protection section collected by airborne lidar at different monitoring times; The classification module 520 is used to determine multiple mileage monitoring units based on the line geographic information, according to the line mileage and facility boundaries, and to determine the candidate object areas corresponding to different object categories within each mileage monitoring unit; The feature point extraction module 530 is used to extract object feature points of the type corresponding to the object category from the line infrastructure objects in the candidate object region according to the object category; The filtering module 540 is used to filter pairs of object feature points with the same name between different monitoring times based on the line mileage, object category, feature point type and local geometric features corresponding to each object feature point; The registration module 550 is used to perform object semantic constraint weighted registration on multi-period point cloud data based on feature point pairs of objects with the same name, according to the object category weight and the matching confidence of each feature point pair of objects with the same name; The monitoring module 560 is used to determine the local deformation data and overall deformation data of the line infrastructure object based on the coordinate differences of the feature points of each object with the same name in the same line infrastructure object after registration, and to determine the deformation monitoring results of the high-speed railway line protection section based on the local deformation data and overall deformation data.

[0141] Figure 6 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 6 As shown, the electronic device may include a processor 610, a communication interface 620, a memory 630, and a communication bus 640. The processor 610, communication interface 620, and memory 630 communicate with each other via the communication bus 640. The processor 610 can call logical instructions from the memory 630 to execute a high-speed rail line protection section monitoring method.

[0142] Furthermore, the logical instructions in the aforementioned memory 630 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0143] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the high-speed railway line protection section monitoring method provided by the above methods.

[0144] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0145] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0146] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for monitoring protected sections of high-speed railway lines, characterized in that, include: Acquire multi-phase point cloud data and line geographic information of the high-speed railway line protection section collected by airborne lidar at different monitoring times; Based on the geographical information of the route, multiple mileage monitoring units are determined according to the route mileage and facility boundaries, and candidate object areas corresponding to different object categories are determined within each of the mileage monitoring units. Based on the object category, extract object feature points of the type corresponding to the object category from the line infrastructure objects in the candidate object region; Based on the line mileage, object category, feature point type and local geometric features corresponding to each object feature point, filter out pairs of object feature points with the same name at different monitoring times; Based on the feature point pairs of the same-named objects, according to the object category weight and the matching confidence of each feature point pair of the same-named objects, the point cloud data of multiple periods are subjected to object semantic constraint weighted registration. Based on the coordinate differences of feature points of the same-named objects in the same line infrastructure object after registration, the local deformation data and overall deformation data of the line infrastructure object are determined, and the deformation monitoring results of the high-speed railway line protection section are determined based on the local deformation data and overall deformation data.

2. The monitoring method for protected sections of high-speed railway lines according to claim 1, characterized in that, Based on the geographical information of the route, multiple mileage monitoring units are determined according to the route mileage and facility boundaries, and candidate object areas corresponding to different object categories are determined within each mileage monitoring unit, including: Based on the geographical information of the line, the point cloud data, the high-speed rail line centerline, mileage information, and the design geometric area of ​​the line infrastructure objects are unified into the same spatial coordinate system. Multiple mileage monitoring units are determined according to the preset mileage length or the facility boundary of the line infrastructure object; Based on the mileage difference between the line mileage corresponding to the point cloud data and the mileage monitoring unit, and the spatial distance between the point cloud data and the design geometric region of the corresponding object category, the candidate object region corresponding to each object category is determined.

3. The monitoring method for protected sections of high-speed railway lines according to claim 1, characterized in that, The object categories include at least one of bridge support objects, slope objects, bridge body objects, and track objects; The step of extracting object feature points of the type corresponding to the object category from the line infrastructure objects in the candidate object region according to the object category includes: When the object category is a bridge support object, extract the end feature points and the contour feature points distributed along the height direction of the bridge support object; When the object category is a slope object, extract the slope boundary feature points, elevation extreme value feature points, and grid sampling feature points of the slope object; When the object category is a bridge main body object, extract the joint feature points and contour corner feature points of the bridge main body object; When the object category is a track object, extract the centerline feature points, joint feature points, and track surface feature points of the track object.

4. The monitoring method for protected sections of high-speed railway lines according to claim 1, characterized in that, The step of filtering pairs of identical object feature points at different monitoring times based on the line mileage, object category, feature point type, and local geometric features corresponding to each object feature point includes: Obtain the line mileage, object category, feature point type, spatial coordinates, normal vector, and curvature corresponding to each of the object's feature points; If two object feature points at different monitoring times have the same object category and the same feature point type, and the corresponding line mileage difference, normal vector difference, and curvature difference satisfy the corresponding preset difference thresholds, the two object feature points are determined as a pair of object feature points with the same name.

5. The monitoring method for protected sections of high-speed railway lines according to claim 4, characterized in that, The step of performing object semantic constraint weighted registration on the point cloud data from multiple periods based on the feature point pairs of the same-named objects, according to the object category weight and the matching confidence of each feature point pair of the same-named objects, includes: Based on the line mileage difference, curvature difference, and normal vector consistency of each pair of feature points of the same object, the matching confidence of each pair of feature points of the same object is determined. The registration error of each pair of object feature points is weighted according to the object category weight corresponding to each object category and the matching confidence of each pair of object feature points with the same name. Based on the weighted registration error, the rotation matrix and translation vector between point cloud data at different monitoring times are determined, and the coordinate transformation of at least one period of point cloud data is performed according to the rotation matrix and translation vector to obtain registered multi-period point cloud data.

6. The monitoring method for protected sections of high-speed railway lines according to claim 1, characterized in that, The determination of the deformation monitoring results of the high-speed railway line protection section based on the local deformation data and the overall deformation data includes: The three-dimensional deformation of each of the same-named object feature points is determined based on the spatial coordinates of each feature point at different monitoring times. Based on the three-dimensional deformation and corresponding stability weight of the feature points of each of the same-named objects in the same line infrastructure object, the overall deformation of the line infrastructure object is determined, and the maximum local deformation of the line infrastructure object is determined based on the three-dimensional deformation of the feature points of each of the same-named objects. The deformation rate is determined based on the overall deformation of the line infrastructure object during continuous monitoring. The risk score of the corresponding line infrastructure object is determined based on the overall deformation, the maximum local deformation, and the deformation rate, and the risk level of the line infrastructure object is determined based on the risk score.

7. The monitoring method for protected sections of high-speed railway lines according to claim 1, characterized in that, Also includes: Point cloud filtering and noise removal are performed on the point cloud data of each period, and point cloud classification, segmentation and geometric feature extraction are performed on the processed point cloud data to obtain the preprocessed point cloud data of each period. The candidate object region is determined based on the corresponding preprocessed point cloud data.

8. A monitoring device for protected sections of high-speed railway lines, characterized in that, include: The information acquisition module is used to acquire multi-phase point cloud data and line geographic information of the high-speed railway line protection section collected by airborne lidar at different monitoring times; The classification module is used to determine multiple mileage monitoring units based on the route geographic information, according to the route mileage and facility boundaries, and to determine candidate object areas corresponding to different object categories within each mileage monitoring unit; The feature point extraction module is used to extract object feature points of the type corresponding to the object category from the line infrastructure objects in the candidate object region according to the object category; The filtering module is used to filter pairs of object feature points with the same name between different monitoring times based on the line mileage, object category, feature point type and local geometric features corresponding to each object feature point; The registration module is used to perform object semantic constraint weighted registration on the point cloud data from multiple periods based on the feature point pairs of the same-named objects, according to the object category weight and the matching confidence of each feature point pair of the same-named objects. The monitoring module is used to determine the local deformation data and overall deformation data of the line infrastructure object based on the coordinate differences of the feature points of each object with the same name in the same line infrastructure object after registration, and to determine the deformation monitoring results of the high-speed railway line protection section based on the local deformation data and overall deformation data.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the high-speed railway line protection section monitoring method as described in any one of claims 1 to 7.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the high-speed railway line protection section monitoring method as described in any one of claims 1 to 7.

Images