Pole tower point cloud and model multi-scale registration method and system based on prior constraint

Abstract

The present application relates to the technical field of three-dimensional image data processing, and more particularly to a tower point cloud and model multi-scale registration method and system based on prior constraints. The method comprises the following steps: S1, standardization processing of multi-source data, so that the point cloud data to be registered and the point cloud representation standard of the prior knowledge base of the candidate model are matched; S2, preliminary screening; S3, coarse registration, through candidate model fast screening and key point priority sampling strategy; S4, fine registration, using soft constraint optimization, quantifying the prior knowledge as weight, and combining weighted ICP and nonlinear least squares solution; S5, according to the matching parameters, the matching score is obtained, and the best matching result of the point cloud data to be registered and the coarse registration candidate model is obtained. A self-learning prior knowledge base is constructed, the high confidence matching result is fed back to update the prior library, the weight strategy and candidate screening rules are dynamically adjusted, and the adaptability of the system to different tower models is improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-dimensional image data processing, and in particular to a tower point cloud and model multi-scale registration method and system based on prior constraints. BACKGROUND

[0002] In the field of three-dimensional modeling and point cloud registration of transmission line towers, existing technologies usually rely on geometric matching or global descriptor methods to align the three-dimensional model of the tower with the measured point cloud. The FBX format of the tower three-dimensional model and the LAS format of the tower point cloud have significant differences in data representation, including coordinate system, unit, scale, attitude and origin position. These heterogeneities make direct registration difficult, and usually require manual adjustment or rely on rough global registration algorithms, such as matching based on point-to-point distance or voxel histogram. However, such methods often have problems of insufficient accuracy and low efficiency in practical applications, making it difficult to meet the high-precision requirements of tower position and structure in transmission line construction and operation.

[0003] On the other hand, due to the noise, uneven point density, occlusion and complex geometric structure of point cloud data, traditional registration methods rely on geometric feature point matching or global descriptor calculation, but these methods are prone to local optimal overall misplacement, especially in the correspondence of key structures (such as tower top, cross arm end point, hanging point, etc.), which has poor stability and is prone to cause high displacement, hierarchical misplacement or cross arm misplacement of the overall model. In addition, existing technologies lack effective use of the topological information and structural features of the tower, such as the number of cross arms, and cannot embed prior knowledge into the registration process.

[0004] Furthermore, existing methods lack adaptive and intelligent mechanisms, and cannot dynamically update the prior library according to historical matching results, nor can they adjust the candidate model selection and matching strategy according to different tower models. Overall, existing technologies have obvious shortcomings in multi-source data compatibility, key structure matching accuracy, automation level, robustness and adaptability, which limits the application and promotion of large-scale tower point cloud and model high-precision registration. SUMMARY

[0005] Therefore, it is necessary to provide a tower point cloud and model multi-scale registration method and system based on prior constraints to solve at least one of the above technical problems.

[0006] To achieve the above purpose, a tower point cloud and model multi-scale registration method based on prior constraints comprises the following steps:

[0007] S1, standardization processing of multi-source data, including:

[0008] S11, constructing a priori knowledge base of the tower three-dimensional model: extracting geometric element information, preset key points, descriptors of the preset key points and semantic attributes of different tower three-dimensional models, and storing point cloud data information corresponding to each tower three-dimensional model into a preset format priori knowledge base;

[0009] S12, tower point cloud data preprocessing: preprocessing the tower point cloud data and extracting global geometric features to construct the point cloud data to be registered; wherein the point cloud data to be registered matches the point cloud representation standard of the priori knowledge base;

[0010] S2, preliminary screening: according to the geometric element information and semantic attributes of the priori knowledge base, the point cloud data to be registered is preliminarily matched, and the tower three-dimensional model with a similarity higher than a preset value is obtained as a candidate model;

[0011] S3, coarse registration: after matching the main shaft of the tower body direction with the Z axis of the point cloud data to be registered, the feature matching of the preset key points in the priori knowledge base is performed under the priori constraint, and the coarse registration result is output, and the candidate model with an error meeting the preset condition is selected as the candidate model for coarse registration;

[0012] S4, fine registration, including:

[0013] S41, assigning different weights to the preset key points in the priori knowledge base and constructing an initial correspondence relationship;

[0014] S42, constructing an alignment target of the point cloud data to be registered and the candidate model for coarse registration by constructing an optimization objective function;

[0015] S43, solving the optimization objective function to obtain the matching parameters of the point cloud data to be registered and each candidate model for coarse registration;

[0016] S5, according to the matching parameters, the matching score is obtained, and the best matching result of the point cloud data to be registered and the candidate model for coarse registration is obtained.

[0017] Preferably, the step S11 specifically includes:

[0018] S111, performing geometric analysis on the tower three-dimensional model, and calculating the geometric element information; the geometric element information includes geometric center, base center, bounding box size and main shaft direction;

[0019] S112, preset key point extraction, including: extraction of global key points, mesoscale key points and local key points; the global key points include tower top and tower bottom, the mesoscale key points include cross arm end points, and the local key points include covering nodes and connecting points;

[0020] S113, description sub-computing: for the extracted global key points, mesoscale key points and local key points, point clouds are generated by grid sampling, and FPFH is used to calculate local descriptors, neighborhood radius and global descriptors;

[0021] S114, extracting semantic attributes: including the mode and information of cross arm layer number, segment number and hanging point;

[0022] S115, statistical analysis of different tower three-dimensional models, extraction of mean value and variance of geometric quantities, construction of corresponding tolerance range, and construction of the prior knowledge base.

[0023] Preferably, the step S2 comprises:

[0024] S21, height constraint: calculating the maximum height difference in the main direction of the point cloud data to be registered and the height of the tower three-dimensional model The height difference , , wherein, is the allowable height deviation, , is the z-axis coordinate value of the i point of the point cloud data to be registered;

[0025] S22, hanging point constraint: calculating the number of conductors and ground wires in the point cloud data to be registered by clustering , according to the number of hanging points in the tower three-dimensional model , ensuring that the number of conductors and ground wires is consistent with the number of hanging points;

[0026] S23, cross arm layer number constraint: the number of tower cross arm layers of the point cloud data to be registered after projection is the same as the number of layers of the tower three-dimensional model ;

[0027] S24, the tower three-dimensional model meeting the constraint conditions of the above steps S21-S23 is taken as the candidate model.

[0028] Preferably, the step S3 comprises:

[0029] S31, obtaining the main direction Z-axis of the point cloud data to be registered by PCA principal component analysis, and rotating the point cloud data to be registered to be consistent with the main axis direction of the candidate model;

[0030] S32, key point matching: calculating the coordinate points in the point cloud data to be registered by clustering, calling the corresponding key point label from the prior knowledge base of the candidate model, and establishing a point-to-point matching relationship between the key points of the point cloud data to be registered and the key points of the candidate model by spatial coordinate comparison;

[0031] S33, according to the preset key point descriptor, using RANSAC random sample consensus algorithm for rough matching, obtaining a tower three-dimensional model with a similarity higher than a preset value as a candidate model, and recording an initial rotation matrix of the point cloud data to be registered relative to the candidate model , a translation vector .

[0032] Preferably, the step S33 includes that the rough matching includes priori knowledge constraint:

[0033] If the distance difference of the hanging points is greater than a set threshold, it is considered that the matching is invalid;

[0034] If the cross arm length difference exceeds the tolerance range, the matching is removed;

[0035] If the number or hierarchical distribution of the preset key points is inconsistent, the error candidate model is directly excluded.

[0036] Preferably, the step S41 includes:

[0037] S411, assigning a differentiated weight to the preset key point i in the priori knowledge base , the preset key point i includes a high-weight key point and a normal point;

[0038] S412, using the initial rotation matrix , a translation vector . The point cloud data to be registered is preliminarily aligned relative to the point cloud of the candidate model;

[0039] S413, key point correspondence construction:

[0040] S4131, for each high-weight key point of the point cloud data to be registered, searching for the most similar candidate point in the point cloud of the candidate model;

[0041] S4132, local descriptor matching: calculating the high-weight key point Fast point feature histogram descriptor, selecting the point with the highest similarity of the descriptor of the high-weight key point and the descriptor of the candidate point as the corresponding point;

[0042] S4133, for each pair of high-weight key points , calculating the matching confidence , ; wherein, is proportional to, is a local descriptor function, is a similarity index, is a key point of the point cloud data to be registered, is a point corresponding to the point cloud of the candidate model;

[0043] S4134, down-sampling each normal point of the point cloud data to be registered to establish an auxiliary correspondence relationship.

[0044] Preferably, the step S42 comprises:

[0045] S421, calculating a point-to-point error for each key point of the point cloud data to be registered : ; wherein, , represents a rotation matrix, is a set of all rotation matrices, used to represent all possible rotations of a rigid body in three-dimensional space; , represents a translation vector, represents a three-dimensional translation vector; S is a set of point indexes participating in optimization;

[0046] S422, constructing a prior constraint loss:

[0047] Tower top and tower bottom center point height alignment: the tower top and tower bottom center points of the point cloud data to be registered and the tower top and tower bottom center points of the candidate model have the minimum height difference, which is calculated by the following formula: ; wherein, represents the Z-axis coordinate of point k of the point cloud data to be registered, represents the Z-axis coordinate of point k of the candidate model;

[0048] Key layer alignment: ensure that the cross arm end points and the hanging points are consistent with the model in the height direction, which is calculated by the following formula: ; wherein, represents the Z-axis coordinate of point j of the point cloud data to be registered, represents the Z-axis coordinate of point j of the candidate model;

[0049] Symmetry and scale constraint: maintain the left-right structure symmetry of the cross arm and the tower body, which is represented by distance constraint as: ; wherein, represents the symmetry distance of point l of the point cloud data to be registered, represents the symmetry distance of point l of the candidate model;

[0050] Form a prior loss by weighted combination: ; wherein, , and are coefficients, adjusted according to the tower type;

[0051] Construct a total optimization objective function: wherein, is a coefficient.​

[0052] Preferably, the step S43 comprises: solving the optimal rigid transformation parameter by using a weighted ICP framework + nonlinear least squares .

[0053] Preferably, the step S5 comprises:

[0054] S51, ICP coincidence degree score : ; N is the number of points participating in registration;

[0055] S52, description sub-consistency score : ; wherein, wherein represents the cosine similarity, wherein, represents the description sub of the point k of the point cloud data to be registered, represents the description sub of the point k of the candidate model; K is the number of point pairs participating in the description sub matching;

[0056] S53, structure consistency score : ; wherein, represents the height difference of the hanging point of the candidate model, represents the height difference of the hanging point of the point cloud data to be registered; M is the number of hanging points participating in the structure feature comparison;

[0057] S54, constructing a comprehensive score function: , alpha, beta, gamma are normalized weight coefficients, controlling the importance of the three indexes (alpha+beta+gamma=1);

[0058] S55, selecting the candidate model of the coarse registration with the highest score as the best matching result.

[0059] The application also provides a tower point cloud and model multi-scale registration system based on prior constraints, comprising a memory, a processor, a computer program stored in the memory and executable on the processor, and the processor executes the computer program to realize the steps of the tower point cloud and model multi-scale registration method based on prior constraints.

[0060] The application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to realize the steps of the tower point cloud and model multi-scale registration method based on prior constraints.

[0061] The tower point cloud and model multi-scale registration method and system based on prior constraints of the application, in view of the problems of multi-source data heterogeneity, point cloud noise interference, key structure matching instability, insufficient utilization of topological information and lack of adaptive mechanism in the prior art, a tower point cloud and model multi-scale registration method based on prior knowledge constraint is proposed. The core goal of the method is to realize high-precision, automation and robustness of tower model and measured point cloud registration. Compared with the prior art, the application has at least the following beneficial effects:

[0062] (1) High matching precision of key structure: By extracting the geometric center, tower top / bottom center point, cross arm end point and node of the tower, and combining local and global descriptors, the geometric, topological and semantic prior information is deeply embedded into the registration process, so as to realize high-precision correspondence of key structures. Compared with the registration method of prior art which only relies on geometric features or global descriptors, the application significantly reduces the risk of tower top offset, cross arm misplacement and overall structure asymmetry, and improves the stability and reliability of the registration result.

[0063] (2) Robust and highly automated registration process: In the coarse registration stage, through candidate model rapid screening and key point priority sampling strategy, the reliability of the initial transformation is improved. In the fine registration stage, soft constraint optimization is adopted, the prior knowledge is quantified as weight, and weighted ICP and nonlinear least squares are combined to solve, so that the registration process can still converge stably in the case of point cloud noise, uneven density or partial occlusion. The whole registration process is highly automated and does not require a lot of manual intervention, greatly improving the registration efficiency and operation convenience.

[0064] (3) Self-learning ability improves system adaptability: With the self-learning mechanism of updating the prior library with high-confidence matching results, the weight strategy and candidate model screening rules can be dynamically adjusted, so as to enhance the adaptability of the system to different tower models and complex environments. Compared with the prior art which lacks adaptive mechanism, the application can continuously optimize the matching precision and robustness, and realize long-term stable operation.

[0065] (4) Comprehensive scoring strategy improves matching reliability: By integrating geometric coincidence, local and global descriptor consistency, structural semantic consistency and other multi-dimensional information, the candidate matching is comprehensively and accurately scored and judged, effectively improving the reliability of the matching result. The prior art relies on a single indicator for evaluation, which may cause the problems of insufficient precision and stability, while the application significantly enhances the comprehensiveness and reliability of the matching judgment.

[0066] (5) The application takes into account registration accuracy, calculation efficiency, automation level and system adaptability, overcomes the deficiencies of prior art in multi-source data processing, key structure matching, prior constraint utilization and self-learning ability, and is especially suitable for high-precision registration of tower point cloud and three-dimensional model and digital twin application in complex environment, and can significantly improve the efficiency and reliability of engineering application. BRIEF DESCRIPTION OF DRAWINGS

[0067] Other features, objects, and advantages of the application will become more apparent from the following detailed description of non-limiting embodiments made with reference to the drawings.

[0068] Figure 1 is a flowchart of a tower point cloud and model multi-scale registration method based on prior constraints provided by the application.

[0069] Figure 2 is a hardware structure schematic diagram of a system running a tower point cloud and model multi-scale registration method based on prior constraints in an embodiment of the application.

[0070] The purpose of the application, functional features and advantages will be further described with reference to the embodiments and the accompanying drawings. DETAILED DESCRIPTION

[0071] The technical method of the application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the application, not all embodiments. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the application.

[0072] In addition, the drawings are only schematic illustrations of the application, and are not necessarily drawn to scale. The same reference signs in the drawings represent the same or similar parts, and therefore repeated description thereof will be omitted. Some block diagrams shown in the drawings are functional entities, which do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or in one or more hardware modules or integrated circuits, or in different network and / or processor methods and / or microcontroller methods.

[0073] To achieve the above-mentioned purpose, please refer to Figure 1 and Figure 2 The application provides a tower point cloud and model multi-scale registration method based on prior constraints, comprising the following steps:

[0074] S1, standardization processing of multi-source data, including:

[0075] S11, constructing a priori knowledge base of the tower three-dimensional model: extracting geometric element information, preset key points, descriptors of the preset key points, and semantic attributes of different tower three-dimensional models, and storing point cloud data information corresponding to each tower three-dimensional model into a preset format priori knowledge base;

[0076] S12, tower point cloud data preprocessing: preprocessing the tower point cloud data and extracting global geometric features to construct the point cloud data to be registered; wherein the point cloud data to be registered matches the point cloud representation standard of the priori knowledge base;

[0077] S2, preliminary screening: according to the geometric element information and semantic attributes of the priori knowledge base, the point cloud data to be registered is preliminarily matched, and the tower three-dimensional model with a similarity higher than a preset value is obtained as a candidate model;

[0078] S3, coarse registration: after matching the main shaft of the tower body direction with the Z-axis of the point cloud data to be registered, the feature matching of the preset key points in the priori knowledge base is performed under the priori constraint, and the coarse registration result is output, and the candidate model with an error meeting the preset condition is selected as the candidate model for coarse registration;

[0079] S4, fine registration, including:

[0080] S41, assigning differentiated weights to the preset key points in the priori knowledge base and constructing an initial correspondence relationship;

[0081] S42, constructing an alignment target of the point cloud data to be registered and the candidate model for coarse registration by constructing an optimization objective function;

[0082] S43, solving the optimization objective function to obtain the matching parameters of the point cloud data to be registered and each candidate model for coarse registration;

[0083] S5, performing matching scoring according to the matching parameters to obtain the best matching result of the point cloud data to be registered and the candidate model for coarse registration.

[0084] Preferably, the step S11 specifically includes:

[0085] S111, performing geometric analysis on the tower three-dimensional model (such as FBX three-dimensional model) to calculate the geometric element information; the geometric element information includes geometric center, base center, bounding box size, and main shaft direction;

[0086] S112, preset key point extraction, including: extraction of global key points, mesoscale key points, and local key points; the global key points include tower top and tower bottom, the mesoscale key points include cross arm end points, and the local key points include covering nodes and connecting points;

[0087] S113, sub-computing: for the extracted global key points, meso-scale key points and local key points, point clouds are generated by grid sampling, and then local descriptors, neighborhood radius and global descriptors are calculated by using fast point feature histogram (FPFH);

[0088] S114, extracting semantic attributes: including the mode and information of cross arm layer number, segment number and hanging point;

[0089] S115, statistical analysis of different tower three-dimensional models is performed to extract the mean and variance of geometric quantities, construct the corresponding tolerance range, and construct the prior knowledge base.

[0090] In a specific example, step S11, the prior knowledge base extraction of the tower three-dimensional model is a key step to construct the prior base, guide point cloud and model registration, and improve matching accuracy and robustness. It specifically includes the following contents:

[0091] 1) geometric meta-information extraction: geometric analysis is performed on the tower three-dimensional model to calculate the geometric center, base center (based on axis-aligned bounding box, AABB), bounding box size and PCA principal axis direction, to provide basic information for model space position, scale and attitude constraint.

[0092] 2) multi-scale key point extraction:

[0093] Global key points: including tower top and tower bottom, used to constrain the overall height range;

[0094] Meso-scale key points: mainly cross arm end points, which can be obtained by horizontal slicing or projection detection of the model;

[0095] Local key points: covering nodes and connection points, which can be obtained by curvature threshold automatic detection or manual annotation.

[0096] The extraction of multi-scale key points ensures that the features of the model at different structural levels can be used for corresponding and constraint.

[0097] 3) geometric and topological descriptor calculation: for the extracted key points, point clouds are first generated by grid sampling, and then local descriptors are calculated by using FPFH. FPFH is calculated for each key point (tower top, tower bottom, cross arm end point, node) using appropriate neighborhood radius (adjusted according to the size of the tower body); at the same time, global descriptors are calculated by using voxel histogram to reflect the overall topological structure. Through the combination of local and global descriptors, multi-level and multi-scale feature expression is formed, which is used for feature matching in subsequent registration.

[0098] 4) Topology and semantic attribute extraction: Extract the topology and semantic attributes of the tower, including the number of cross arms, the number of sections, and the hanging point mode. This information can be obtained through manual annotation or automatically identified using a deep learning segmentation network to enhance the semantic constraint ability during registration. For example, it can be obtained through manual annotation.

[0099] 5) Statistical template and index construction: Perform statistical analysis on similar model towers to extract the mean and variance of geometric quantities such as cross arm spacing and layer height, and construct the corresponding tolerance range. Store the above geometric meta-information, key points, descriptors, and semantic attributes as a prior knowledge base in JSON format for quick retrieval and invocation.

[0100] Through the above steps, a complete FBX model prior knowledge base is formed, providing reliable geometric, topological, and semantic constraints for subsequent tower point cloud and model multi-scale registration.

[0101] Preferably, the step S2 comprises:

[0102] S21, height constraint: calculate the maximum height difference in the main direction of the point cloud data to be registered and the height of the tower three-dimensional model The height difference , wherein, is the allowed height deviation (e.g., 5%), , is the z-axis coordinate value of the i-th point of the point cloud data to be registered;

[0103] S22, hanging point constraint: calculate the number of conductors and ground wires in the point cloud data to be registered by clustering According to the number of hanging points in the tower three-dimensional model Ensure that the number of conductors and ground wires is consistent with the number of hanging points;

[0104] S23, cross arm layer number constraint: the number of cross arm layers of the tower body after projection of the point cloud data to be registered is the same as the number of layers of the tower three-dimensional model ;

[0105] S24, the tower three-dimensional model that meets the constraint conditions of the above steps S21-S23 is taken as the candidate model.

[0106] Preferably, the step S3 comprises:

[0107] S31, obtain the main direction Z-axis of the point cloud data to be registered through PCA principal component analysis, and rotate the point cloud data to be registered to be consistent with the main axis direction of the candidate model;

[0108] S32, key point matching: calculate the coordinate points in the point cloud data to be registered, call the corresponding key point label from the prior knowledge base of the candidate model, preliminarily correspond the key points of the point cloud data to be registered and the key points of the candidate model through spatial coordinate comparison, and establish a point-to-point matching relationship;

[0109] S33, according to the descriptor of the preset key point, using RANSAC random sample consensus algorithm for coarse matching, obtaining a tower three-dimensional model with a similarity higher than a preset value as a candidate model, and recording an initial rotation matrix of the point cloud data to be registered relative to the candidate model , translation vector .

[0110] Preferably, the step S33 includes that the coarse matching includes prior knowledge constraint:

[0111] If the distance difference of the hanging points is greater than the set threshold, it is considered that the matching is invalid;

[0112] If the cross arm length difference exceeds the tolerance range, the matching is excluded;

[0113] If the number or hierarchical distribution of the preset key points is inconsistent, the error candidate model is directly excluded.

[0114] Before fine registration, coarse registration needs to be completed first to ensure that the point cloud and the tower model point cloud are roughly aligned in overall position, size and direction. The goal of coarse registration is to preliminarily constrain the point cloud by using prior knowledge, to narrow the search space and provide a reliable initial solution for subsequent fine registration. In a specific example, step S3 can include:

[0115] 1) input:

[0116] LAS point cloud: actual tower point cloud data to be registered;

[0117] Candidate FBX model point cloud and its prior knowledge.

[0118] 2) initial rotation based on PCA:

[0119] The main direction (generally the tower body direction) of the point cloud has been obtained by PCA (principal component analysis), and the rotation angle of the point cloud around the Z axis can be calculated. The point cloud is rotated to be consistent with the main axis direction of the FBX model, so as to eliminate large-scale rotation uncertainty. This step can significantly reduce the calculation amount in subsequent matching and improve the stability of matching convergence.

[0120] 3) key point matching: extract key points from the point cloud:

[0121] Hanging point extraction: By DBSCAN clustering of tower point cloud and combining distance constraints, the nearest point corresponding to each conductor and ground wire is selected as the hanging point.

[0122] Tower top point extraction: The point cloud height maximum point is taken, and the x, y distance from the tower center is combined to exclude abnormal points.

[0123] Cross arm end point: obtained by clustering and edge detection.

[0124] From the FBX model prior library, the corresponding key point label is called, and through spatial coordinate comparison, the point cloud key point and the model key point are preliminarily corresponding, and a point-to-point matching relationship is established, providing candidate pairs for subsequent RANSAC coarse registration.

[0125] 4) Feature matching under prior constraints:

[0126] Extract the local feature descriptor FPFH of the point cloud, and use the RANSAC random sample consensus algorithm for coarse matching. In the matching process, introduce prior knowledge constraints:

[0127] If the distance difference of the hanging point is greater than the set threshold, it is considered that the matching is invalid;

[0128] If the cross arm length difference is too large (out of tolerance range), the matching is excluded;

[0129] If the number of key points or hierarchical distribution is inconsistent, the wrong candidate model is directly excluded.

[0130] 5) Coarse registration output result:

[0131] Align the point cloud and the FBX model point cloud in the overall rotation, scale and position, and select the five candidate matching results with the smallest error according to the prior knowledge as the initial input of fine registration, so as to ensure that the subsequent optimization process converges faster and the result is more stable.

[0132] Preferably, the step S41 comprises:

[0133] S411, assigning different weights to the preset key points i in the prior knowledge base , the preset key points i include high-weight key points and ordinary points;

[0134] S412, using the initial rotation matrix , translation vector output by coarse registration. Preliminarily align the point cloud data to be registered with the point cloud of the candidate model;

[0135] S413, key point correspondence construction:

[0136] S4131, for each high-weight key point of the point cloud data to be registered, searching for the most similar candidate point in the point cloud of the candidate model;

[0137] S4132, local descriptor matching: calculating the high-weight key point Fast Point Feature Histogram (FPFH) descriptor, selecting the point with the highest similarity between the descriptor of the high-weight key point and the descriptor of the candidate point as the corresponding point;

[0138] S4133, for each pair of high-weight key points calculating the matching confidence , ; wherein, is proportional to, represents a local descriptor function, is a similarity index, represents a key point of the point cloud data to be registered, represents the corresponding point in the point cloud of the candidate model; : the point in the FBX candidate model point cloud;

[0139] S4134, for each normal point of the point cloud data to be registered, performing downsampling processing to establish an auxiliary correspondence relationship.

[0140] Preferably, the step S42 comprises:

[0141] S421, for each key point of the point cloud data to be registered , calculating the point-to-point error : ; wherein, , represents a rotation matrix, is a set of all rotation matrices; , represents a translation vector, represents a three-dimensional translation vector; S is a set of point indexes participating in optimization;

[0142] S422, constructing a prior constraint loss:

[0143] Tower top and tower bottom center point height alignment: the tower top and tower bottom center points of the point cloud data to be registered and the tower top and tower bottom center points of the candidate model have the minimum height difference, which is calculated by the following formula: ; wherein, represents the Z-axis coordinate of point k of the point cloud data to be registered, represents the Z-axis coordinate of point k of the candidate model;

[0144] Key layer alignment: ensure the cross arm end point and the hanging point are consistent with the model in the height direction, calculated by the following formula: ; wherein, represents the Z-axis coordinate of point j of the point cloud data to be registered, represents the Z-axis coordinate of point j of the candidate model;

[0145] Symmetry and scale constraint: keep the cross arm and tower body structure symmetric on the left and right, expressed by distance constraint as: ; wherein,

[0146] represents the symmetry distance of point l of the point cloud data to be registered, represents the symmetry distance of point l of the candidate model;

[0147] Form the prior loss by weighted combination: ; wherein, , and are coefficients, adjusted according to the tower type;

[0148] Construct the total optimization objective function: , wherein, is a coefficient.

[0149] Preferably, the step S43 comprises: solving the optimal rigid body transformation parameter by using the weighted ICP framework + nonlinear least squares .

[0150] Preliminary alignment of point cloud and FBX model point cloud in overall rotation, scale and position, and selection of candidate matching results with smaller error (for example, the 5 results with the smallest error, or the results with error less than a certain threshold) according to prior knowledge as the initial input of fine registration, so as to ensure that the subsequent optimization process converges faster and the result is more stable. In a specific example, the fine registration of step S4 can include:

[0151] 1) input:

[0152] LAS point cloud: actual tower point cloud data to be registered;

[0153] Coarse registration candidate FBX model point cloud and its prior information;

[0154] Coarse registration output: initial rotation matrix , translation vector .

[0155] 2) Key point weight and matching construction:

[0156] In the fine registration stage, to ensure the high-precision alignment of the tower point cloud and the model at key structural parts, differentiated weights need to be assigned to the key points, and an initial correspondence relationship needs to be established.

[0157] (a) Key point weight distribution:

[0158] According to the importance of the tower structure, the points in the point cloud are divided into two categories: key points and ordinary points:

[0159] a) Key points (high weight): including: tower top, cross arm end point, hanging point; Assign a higher weight to ensure that these points dominate the estimation of rigid body transformation in the optimization objective; The weight can be dynamically adjusted according to the importance of the point or the description of the sub-matching confidence, for example: ;

[0160] b) Ordinary points (low weight): non-key points on the tower, assigned a lower weight to provide overall geometric information, but not to dominate the optimization: ;

[0161] Through this differentiated weight distribution, the key structural points can dominate the correction of rotation and translation in the optimization process, thereby improving the matching accuracy of the tower top, cross arm and node.

[0162] (b) Refine point correspondence based on coarse registration:

[0163] a) Initial alignment input: use the initial rigid body transformation output by coarse registration , to roughly align the LAS point cloud with the FBX model point cloud. This initial alignment ensures that the point cloud and the model are close in overall rotation, scale and position, facilitating the subsequent refinement of the correspondence relationship.

[0164] b) Key point correspondence construction: precise matching for structural key points (tower top, cross arm end point, hanging point, etc.):

[0165] For each LAS point cloud key point , search for the most similar candidate point in the model point cloud ;

[0166] Local descriptor matching: calculate the FPFH descriptor and select the point with the highest descriptor similarity as the correspondence;

[0167] Calculate the matching confidence for each pair of key points , and the point pair with high confidence has a greater weight in the subsequent optimization: ; where is proportional to , is the local descriptor function, and is the similarity index.

[0168] c) General point correspondence construction: For non-key points, down-sampling is performed first to establish auxiliary correspondence to provide overall geometric constraints: nearest neighbor sampling matching. Assign a lower weight to ensure that the key structure is not disturbed too much during optimization, while also helping to reduce overall shape error.

[0169] 3) Construction of optimization objective: The core of precise registration is to achieve high-precision alignment of LAS point cloud and candidate FBX model in geometry, topology and semantics through optimization objective function. The optimization objective is usually composed of two parts: point-to-point weighted error and prior constraint loss.

[0170] a) Point-to-point weighted error: for each point cloud key point (from the LAS point cloud), assuming its corresponding FBX model point is , introduce the weighted coefficient to represent the credibility or importance of each point (key points such as tower top, tower bottom center, cross arm end point, hanging point can be set with higher weight). Then the point-to-point error is : ; where , represents the rotation matrix, is the set of all rotation matrices, which represents all possible rotations of the rigid body in three-dimensional space; , represents the translation vector, represents the three-dimensional translation vector; S is the set of point indices participating in optimization; : is the corresponding index of point in the FBX model point cloud;

[0171] This term ensures that the point cloud geometry and the model coincide as much as possible, and is the basis of ICP.

[0172] b) Prior constraint loss: In addition to point-to-point error, precise registration needs to combine prior knowledge of tower structure to increase geometric and topological constraints, making the result more consistent with the real tower shape. The constraints include:

[0173] (a) Tower top / tower bottom center point height alignment: ensure that the tower top and tower bottom center points of the LAS point cloud have the smallest height difference with the model tower top and tower bottom center points, calculated by the following formula:

[0174] (b) Key layer alignment: ensure that the cross arm end points and hanging points are consistent with the model in the height direction, calculated by the following formula: ;

[0175] (c) Symmetry and scale constraint: maintain the symmetry of the left and right structures of the cross arm and tower body, represented by distance constraint: ;

[0176] These constraints are combined by weighting to form the prior loss: ; where, , and are coefficients adjusted according to the tower type;

[0177] c) Comprehensive optimization goal: combine the point-to-point error with the prior constraint to obtain the total optimization goal function:

[0178] ; this goal function takes into account both the point cloud geometric coincidence and the tower prior structure constraint, and is the core of the precise registration solution.

[0179] 4) Optimization solution: adopt the weighted ICP framework + nonlinear least squares, that is, put the point-to-point residual error and the prior constraint residual error into an overall optimization goal, balance them with weights, and then solve the optimal rigid body transformation parameters through nonlinear least squares .

[0180] The skilled person can design the optimization solution as needed, and the present application will not be repeated. For example, the iterative step can include: updating the point correspondence ; calculating the point-to-point residual error and the prior constraint residual error; minimizing the objective function to update the rotation and translation ; convergence condition: error reduction is less than threshold or maximum iteration number is reached.

[0181] 5) Output: after fine registration, output the relevant parameters of each model to provide data support for subsequent matching scoring.

[0182] Preferably, the step S5 comprises:

[0183] S51, ICP coincidence score : ; N is the number of points participating in registration;

[0184] S52, description sub-consistency score : ; where, where represents the cosine similarity, where, represents the descriptor of point k of the point cloud data to be registered, represents the descriptor of point k of the candidate model; K is the number of point pairs participating in descriptor matching;

[0185] S53, structure consistency score : ; where, represents the height difference of the hanging point of the candidate model, height difference of the hanging points representing the point cloud data to be registered; M is the number of hanging points participating in the comparison of the structural features;

[0186] S54, constructing a comprehensive score function: , α, β, γ are normalized weight coefficients, controlling the importance of the three indexes (α+β+γ=1);

[0187] S55, selecting the candidate model of the coarse registration with the highest score as the best matching result.

[0188] The application further provides a tower point cloud and model multi-scale registration system based on prior constraints, which is run on the basis of a computer system and specifically includes a memory 61, a processor 62, and a computer program 63 stored in the memory 61 and capable of running on the processor 62, wherein the processor 62 implements the steps of the tower point cloud and model multi-scale registration method based on prior constraints as described above when the computer program 63 is executed.

[0189] Illustratively, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the application. The one or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the computer.

[0190] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor.

[0191] The memory can be an internal storage unit, such as a hard disk or a memory; the memory can also be an external storage device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. Further, the memory can also include both an internal storage unit and an external storage device. The memory is used to store the computer program and other programs and data required by the terminal device. The memory can also be used to temporarily store data that has been output or will be output.

[0192] The application further provides a computer readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the tower point cloud and model multi-scale registration method based on prior constraints as described above.

[0193] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above functional units and modules is exemplified, and in actual application, the above functions can be completed by different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated in one processing unit, or each unit can exist physically, or two or more units can be integrated in one unit. The integrated unit can be realized in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for easy distinction, and do not limit the protection scope of the application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiments, which will not be described here.

[0194] In the above embodiments, the description of each embodiment has its own emphasis, and the parts not described or recorded in detail in a certain embodiment can be referred to the relevant description of other embodiments.

[0195] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be realized in electronic hardware or a combination of computer software and electronic hardware. Whether the functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the application.

[0196] In the embodiments of the present application, it should be understood that the disclosed apparatus / system and method can be implemented in other manners. For example, the embodiments of the apparatus / terminal device described above are merely schematic, and the division of the modules or units is merely a logical function division, and there can be another division manner in actual implementation. For example, a plurality of units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed mutual couplings or direct couplings or communication connections between different units, can be indirect couplings or communication connections through some interfaces, devices or units, and can be electrical, mechanical or in other forms.

[0197] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units, i.e., can be located in one place, or can be distributed on a plurality of network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the embodiments.

[0198] In addition, each functional unit in the various embodiments of the present application can be integrated into a processing unit, or each unit can be a physically independent unit, or two or more units can be integrated into a unit. The integrated unit can be implemented in the form of hardware or in the form of a software functional unit.

[0199] The integrated module / unit, if implemented in the form of a software functional unit and sold or used as an independent product, can be stored in a computer readable storage medium. Based on such understanding, all or part of the flow of the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the computer program can be stored in a computer readable storage medium. When the processor executes the computer program, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable file or some intermediate form. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the computer readable medium can include appropriate contents according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0200] Thus, the embodiments should be considered in all respects as illustrative and not restrictive, the scope of the application being indicated by the appended claims rather than by the description given above, and all changes which come within the meaning and range of equivalency of the claims are intended to be embraced therein.

[0201] The foregoing is considered as illustrative only of the principles of the application. Numerous modifications and changes will readily occur to those skilled in the art, and the generic principles defined herein can be applied to other embodiments without departing from the spirit or scope of the application. Therefore, the application is not intended to be limited to the embodiments shown herein but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for multi-scale registration of tower point clouds and models based on prior constraints, characterized in that, Includes the following steps: S1, standardization processing of multi-source data, including: S11, Construction of the prior knowledge base for the 3D model of the pole: Extract the geometric element information, preset key points, descriptors and semantic attributes of the preset key points of different 3D models of poles, and store the point cloud data information corresponding to each 3D model of poles as a prior knowledge base in a preset format; the semantic attributes include the number of crossarm layers, the number of segments and the pattern and information of the connection points; S12, Tower point cloud data preprocessing: The tower point cloud data is preprocessed and global geometric features are extracted to construct point cloud data to be registered; wherein, the point cloud data to be registered is matched with the point cloud representation standard of the prior knowledge base; S2, Preliminary screening: Based on the geometric element information and semantic attributes of the prior knowledge base, the point cloud data to be registered is initially matched, and the 3D models of the tower with similarity higher than the preset value are obtained as candidate models. S3, Coarse Registration: After matching the main axis of the tower body direction with the Z-axis of the point cloud data to be registered, feature matching under prior constraints is performed on the preset key points in the prior knowledge base, and coarse registration results are output. Candidate models whose errors meet the preset conditions are selected as candidate models for coarse registration. S4, fine registration, includes: S41, assign differentiated weights to preset key points in the prior knowledge base and construct an initial correspondence; S42, by constructing an optimized objective function, an alignment target is built between the point cloud data to be registered and the candidate model for coarse registration; S43, Solve the optimization objective function to obtain the matching parameters between the point cloud data to be registered and each candidate model for coarse registration; specifically, use the weighted ICP framework + nonlinear least squares to solve for the optimal rigid body transformation parameters; S5, perform matching scoring based on the matching parameters, specifically including: scoring and judging candidate matches by integrating multi-dimensional information such as geometric overlap, local and global descriptor consistency, and structural semantic consistency; and obtaining the best matching result between the point cloud data to be registered and the candidate model of coarse registration.

2. The method for multi-scale registration of tower point clouds and models based on prior constraints according to claim 1, characterized in that, Step S11 specifically includes: S111, Perform geometric analysis on the three-dimensional model of the tower and calculate the geometric element information; the geometric element information includes the geometric center, the base center, the boundary box dimensions, and the principal axis direction; S112, Preset key point extraction, including: extraction of global key points, mesoscale key points and local key points; the global key points include the top and bottom of the tower, the mesoscale key points include the crossarm endpoints, and the local key points include the cover nodes and connection points. S113, Descriptor Calculation: For the extracted global key points, mesoscale key points, and local key points, a point cloud is generated through grid sampling, and then the local descriptor, neighborhood radius, and global descriptor are calculated using FPFH. S114, Extract semantic attributes: including the number of crossarm layers, number of segments, and the pattern and information of the connection points; S115, Perform statistical analysis on different tower 3D models, extract the mean and variance of geometric quantities, construct the corresponding tolerance range, and construct the prior knowledge base.

3. The method for multi-scale registration of tower point clouds and models based on prior constraints according to claim 2, characterized in that, Step S2 includes: S21, Height Constraint: Calculate the maximum height difference in the principal direction of the point cloud data to be registered. The height of the three-dimensional model of the tower. height difference , ,in, To allow for height deviation, , The z-axis coordinate of point i in the point cloud data to be registered; S22, Connection Point Constraint: Calculate the number of wires and ground wires in the point cloud data to be registered using clustering. Based on the number of attachment points in the 3D model of the tower Ensure that the number of conductors and ground wires matches the number of connection points; S23, Crossarm Count Constraint: After projecting the point cloud data to be registered, the number of crossarms in the tower body is counted. The number of layers in the 3D model of the tower same; S24. Select the three-dimensional model of the tower that meets the constraints of steps S21 to S23 above as the candidate model.

4. The method for multi-scale registration of tower point clouds and models based on prior constraints according to claim 3, characterized in that, Step S3 includes: S31, obtain the principal direction Z-axis of the point cloud data to be registered through PCA principal component analysis, and rotate the point cloud data to be registered to be consistent with the principal axis direction of the candidate model; S32, Key point matching: The coordinates of the points in the point cloud data to be registered are calculated by clustering. The corresponding key point annotations are called from the prior knowledge base of the candidate model. By comparing the spatial coordinates, the key points of the point cloud data to be registered are initially matched with the key points of the candidate model, and a point-to-point matching relationship is established. S33, based on the descriptors of the preset key points, a coarse matching is performed using the RANSAC random sampling consensus algorithm to obtain 3D tower models with similarity higher than a preset value as candidate models, and the initial rotation matrix of the point cloud data to be registered relative to the candidate models is recorded. Translation vector .

5. The method for multi-scale registration of tower point clouds and models based on prior constraints according to claim 4, characterized in that, Step S33 includes the coarse matching including prior knowledge constraints: If the distance difference between the hook points is greater than a set threshold, the match is considered invalid. If the difference in crossarm length exceeds the tolerance range, the match is rejected. If the preset number of key points or the hierarchical distribution are inconsistent, incorrect candidate models will be directly excluded.

6. The method for multi-scale registration of tower point clouds and models based on prior constraints according to claim 4, characterized in that, Step S41 includes: S411, assign differentiated weights to the preset key points i in the prior knowledge base. The preset key point i includes high-weight key points and ordinary points; S412, using the initial rotation matrix output from coarse registration. Translation vector The point cloud data to be registered is initially aligned with the point cloud of the candidate model. S413, Key Point Correspondence Construction: S4131: For each high-weight key point in the point cloud data to be registered, search for the most similar candidate point in the point cloud of the candidate model. S4132, Local Descriptor Matching: Calculating High-Weight Keypoints Fast point feature histogram descriptors: select the point with the highest similarity between the descriptor of the high-weight key point and the descriptor of the candidate point as the corresponding point; S4133, for each pair of high-weight key points Calculate the matching confidence score , ;in, Indicates proportional to, This represents a local descriptor function. As a similarity index, This indicates the key points of the point cloud data to be registered. express The corresponding points in the point cloud of the candidate model; S4134 performs downsampling processing on each ordinary point in the point cloud data to be registered, and establishes an auxiliary correspondence.

7. The method for multi-scale registration of tower point clouds and models based on prior constraints according to claim 6, characterized in that, Step S42 includes: S421, each key point in the point cloud data to be registered Calculate point-to-point error : ;in, , Represents the rotation matrix. Let be the set of all rotation matrices; , Represents the translation vector. S represents the three-dimensional translation vector; S is the set of point indices participating in the optimization. S422, Constructing the prior constraint loss: Alignment of the center points of the tower top and bottom: The height difference between the center points of the tower top and bottom of the point cloud data to be registered and the center points of the tower top and bottom of the candidate model is minimized, calculated by the following formula: ;in, This represents the Z-axis coordinate of point k in the point cloud data to be registered. The Z-axis coordinate of point k in the candidate model; Key layer alignment: Ensure that the crossarm endpoints and attachment points are aligned with the model in the height direction, calculated using the following formula: ;in, This represents the Z-axis coordinate of point j in the point cloud data to be registered. Represents the Z-axis coordinate of point j in the candidate model; Symmetry and Proportion Constraints: Maintaining the symmetry of the crossarm and the left and right structures of the tower is expressed by distance constraints as follows: ;in, This represents the symmetric distance of point l in the point cloud data to be registered. Represents the symmetric distance of point l in the candidate model; The prior loss is formed by weighted combination: ;in, , as well as This is a coefficient, adjusted according to the tower type; Construct the overall optimization objective function: ,in, is a coefficient.

8. The method for multi-scale registration of tower point clouds and models based on prior constraints according to claim 7, characterized in that, Step S43 includes: using a weighted ICP framework + nonlinear least squares to solve for the optimal rigid body transformation parameters. .

9. The method for multi-scale registration of tower point clouds and models based on prior constraints according to claim 8, characterized in that, Step S5 includes: S51, ICP overlap score : N is the number of points involved in the registration. S52, Descriptor Consistency Score : Among them, Represents cosine similarity, where, A descriptor representing point k in the point cloud data to be registered. The descriptor of point k in the candidate model; K is the number of point pairs participating in the descriptor matching; S53, Structural Consistency Score : ;in, The height difference of the attachment points representing the candidate models. The height difference of the attachment points in the point cloud data to be registered represents the height difference of the attachment points; M is the number of attachment points participating in the structural feature comparison. S54, Construct the comprehensive scoring function: α, β, γ are normalized weighting coefficients, and the importance of the three indicators is controlled by α + β + γ = 1. S55, select the candidate model with the highest coarse registration score as the best matching result.

10. A multi-scale registration system for pole point clouds and models based on prior constraints, 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 computer program, it implements the steps of the method for multi-scale registration of tower point cloud and model based on prior constraints as described in any one of claims 1 to 9.

Images