Detection method and device based on point cloud compression and cascade rendering

Abstract

The invention discloses a detection method and device based on point cloud compression and cascade rendering. The method comprises the steps of obtaining point cloud data of a construction site and a building informatization model; determining a construction scene according to the building informatization model; compressing the point cloud data according to the construction scene and a predetermined scene-compression matching strategy; rendering the compressed point cloud data according to a construction scene and a predetermined scene-rendering matching strategy to generate a three-dimensional rendering graph; and carrying out difference comparison on the building informatization model and the three-dimensional rendering graph to obtain a construction difference. According to the method, the point cloud data processing and rendering efficiency can be effectively improved in a complex construction environment, and the difference between a construction site and a design scheme is detected.

Description

Technical Field

[0001] The present invention belongs to the technical field of electric power construction, and in particular relates to a detection method and device based on point cloud compression and cascade rendering. Background Art

[0002] With the accelerated advancement of the construction of new power systems, the scale and complexity of power projects have reached a historical high. In this context, high-precision real-time monitoring of the construction process has become a core proposition to ensure project safety and control construction costs. Although point cloud technology has been widely used in the field of power engineering, the contradiction between its massive data characteristics and the adaptability of traditional construction monitoring systems has become increasingly prominent. At least the following problems still exist: 1. Processing difficulties under data floods: In modern power engineering, the amount of point cloud data generated by a single aerial survey can reach tens of GB, and the three-dimensional modeling of ultra-high voltage converter stations requires the integration of more than 2 million measurement points. Traditional data processing processes face challenges such as surging storage costs, computing resource bottlenecks, and low data utilization. 2. The timeliness of static comparison technology has failed: The current BIM+point cloud construction comparison technology still relies on manual inspections and offline analysis, which is difficult to meet dynamic construction needs. There are problems such as lagging progress control, difficult quality traceability, and inefficient collaborative operations. For example, CN115540752A discloses a system and construction method for detecting allowable deviations in engineering quality based on laser point cloud, wherein the detection system includes a user operation interface module, a three-dimensional rendering module, a plane detection module and a measurement and calculation module; the user operation interface module is used to provide a user operation interface and a graphical interactive interface, specifically including: file import, display of rendering model, display of model block list, provision of plane list and output of measurement information; the three-dimensional rendering module is used for point cloud segmentation, rendering and highlighting; the plane detection module is used to extract plane information of imported point cloud data; the measurement and calculation module is used to calculate the numerical results of quality inspection items.

[0003] Therefore, how to provide a detection method that can effectively improve data processing efficiency and achieve difference comparison is a technical problem that needs to be solved urgently in this field. Summary of the invention

[0004] In view of the defects in the above-mentioned prior art, the present invention provides a detection method and device based on point cloud compression and cascade rendering. The present invention can effectively improve the efficiency of point cloud data processing and rendering in a complex construction environment, and detect the difference between the construction site and the design plan.

[0005] In a first aspect, the present invention provides a detection method based on point cloud compression and cascade rendering, comprising:

[0006] Obtain point cloud data and building information models of the construction site;

[0007] Determine the construction scenario based on the building information model;

[0008] Compress the point cloud data according to the construction scenario and the pre-determined scenario-compression matching strategy;

[0009] Render the compressed point cloud data according to the construction scenario and the pre-determined scenario-rendering matching strategy to generate a three-dimensional rendering;

[0010] Compare the building information model with the three-dimensional rendering to obtain construction differences.

[0011] Furthermore, determine the construction scenario according to the building information model, including:

[0012] Determine the spatial information and process flow according to the building information model;

[0013] Traverse the equipment types involved in the process flow to determine whether there is a unique equipment corresponding to a certain construction scenario;

[0014] When there is no unique equipment corresponding to a certain construction scenario in the process flow, match the equipment type and spatial information with the scenario characteristics of each construction scenario, and give the corresponding matched construction scenario.

[0015] Furthermore, match the equipment type and spatial information with the scenario characteristics of each construction scenario, and give the corresponding matched construction scenario, including:

[0016] Obtain the necessary equipment types and spatial limitation conditions for each construction scenario;

[0017] Match the equipment type with the necessary equipment types of each construction scenario to give a preliminary matched construction scenario;

[0018] Match the spatial information with the spatial limitation conditions corresponding to the preliminary matched construction scenario to give the construction scenario corresponding to the spatial limitation conditions satisfied by the spatial information;

[0019] Based on the construction scenario corresponding to the spatial limitation conditions satisfied by the spatial information, give the corresponding scoring rules and score the spatial information to determine the construction scenario with the highest score.

[0020] Furthermore, the pre-determination of the scenario-compression matching strategy includes:

[0021] Obtain the historical point cloud data corresponding to different construction scenarios and various point cloud compression algorithms;

[0022] Compress the historical point cloud data corresponding to each construction scenario respectively using various point cloud compression algorithms;

[0023] Calculate the data compression ratio and point cloud geometric deviation between the historical point cloud data compressed by different point cloud compression algorithms for all construction scenarios and the historical point cloud data before its own compression;

[0024] Determine the best point cloud compression algorithm for each construction scenario according to the data compression ratio and point cloud geometric deviation corresponding to the historical point cloud data compressed by different point cloud compression algorithms for each construction scenario;

[0025] Associate each construction scenario with the determined best point cloud compression algorithm.

[0026] Furthermore, determining the best point cloud compression algorithm for each construction scenario according to the data compression ratio and point cloud geometric deviation corresponding to the historical point cloud data compressed by different point cloud compression algorithms for each construction scenario includes:

[0027] Conduct a comprehensive analysis of the data compression ratio and point cloud geometric deviation to determine the compression score corresponding to the historical point cloud data compressed by different point cloud compression algorithms for each construction scenario;

[0028] Select the point cloud compression algorithm corresponding to the highest compression score in each construction scenario;

[0029] Among them, the compression score satisfies the following relationship:

[0030]

[0031] In the formula, S is the compression score, α is the weight of the data compression ratio, β is the weight of the point cloud geometric deviation, CR is the data compression ratio, and PC is the point cloud geometric deviation.

[0032] Furthermore, the pre-determination of the scene-rendering matching strategy includes:

[0033] Obtain the historical point cloud data compressed by the pre-determined scene-compression matching strategy for different construction scenarios and multiple cascaded rendering algorithms;

[0034] Render the historical point cloud data corresponding to each construction scenario using multiple cascaded rendering algorithms respectively to obtain historical 3D rendered images;

[0035] Calculate the rendering scores for all historical 3D rendered images and give the cascaded rendering algorithm with the highest rendering score corresponding to each construction scenario;

[0036] Associate each construction scenario with the cascaded rendering algorithm with the highest rendering score.

[0037] Furthermore, calculating the rendering scores for all historical 3D rendered images includes:

[0038] Obtain the key devices and their corresponding spatial positions in each historical 3D rendering;

[0039] Determine the visual light and shadow difference degree based on each historical 3D rendering and the actual construction site image corresponding to the historical point cloud data;

[0040] Determine the detail difference degree based on each key device in each historical 3D rendering and the corresponding device before the compression of the historical point cloud data;

[0041] Determine the position deviation degree based on the spatial position of each key device in each historical 3D rendering and the spatial position before rendering;

[0042] Comprehensively analyze the visual light and shadow difference degree, detail difference degree, and position deviation degree of each historical 3D rendering respectively, and give the rendering score corresponding to each historical 3D rendering.

[0043] Furthermore, determining the visual light and shadow difference degree based on each historical 3D rendering and the actual construction site image corresponding to the historical point cloud data includes:

[0044] Extract the lighting parameters and perform color conversion on the historical 3D rendering and the actual construction site image respectively, and calculate the brightness gradient change of the historical 3D rendering;

[0045] Compare the lighting parameters of the historical 3D rendering and the actual construction site image, and give the shadow matching degree;

[0046] Compare the color conversion results of the historical 3D rendering and the actual construction site image, and give the color difference;

[0047] Comprehensively analyze the color difference and the calculation result of the brightness gradient change after normalization and the shadow matching degree respectively, and give the visual light and shadow matching calculation result.

[0048] Furthermore, determining the detail difference based on each key device in each historical 3D rendering and the corresponding device before the compression of the historical point cloud data includes:

[0049] Perform edge detection analysis and curvature change detection analysis on each historical 3D rendering respectively, and give the edge sharpness and surface fidelity;

[0050] Perform geometric shape matching analysis on each key device in each historical 3D rendering and the corresponding device before the compression of the historical point cloud data, and give the shape matching degree;

[0051] Comprehensively analyze the shape matching degree, edge sharpness, and surface fidelity, and give the detail difference result.

[0052] Further, compress the point cloud data according to the construction scenario and the pre-determined scenario-compression matching strategy, including:

[0053] Based on the pre-determined scenario-compression matching strategy and the construction scenario, determine the optimal point cloud compression algorithm associated with the construction scenario;

[0054] Compress the point cloud data according to the determined optimal point cloud compression algorithm.

[0055] Further, render the compressed point cloud data according to the construction scenario and the pre-determined scenario-rendering matching strategy to generate a three-dimensional rendering, including:

[0056] Based on the construction scenario, determine the cascaded rendering algorithm associated with the construction scenario;

[0057] Render the compressed point cloud data according to the determined cascaded rendering algorithm to generate a three-dimensional rendering.

[0058] Further, compare the differences between the building information model and the three-dimensional rendering to obtain construction differences, including:

[0059] Extract the features of the building information model and the three-dimensional rendering respectively, and give their respective key feature points;

[0060] Match the key feature points of the building information model and the three-dimensional rendering, and give the set of matched key feature point pairs;

[0061] Based on the set of matched key feature point pairs, determine the rigid transformation matrix and align the building information model and the three-dimensional rendering;

[0062] Based on the rigid transformation matrix, perform point-by-point calculations on the aligned building information model and three-dimensional rendering, and give the offset and missing area.

[0063] Further, based on the rigid transformation matrix, perform point-by-point calculations on the aligned building information model and three-dimensional rendering, and give the offset and missing area, including:

[0064] Based on the rigid transformation matrix, determine the distance error of each corresponding point position in the building information model and the three-dimensional rendering;

[0065] Based on the distance error of each point position and the preset distance error threshold, determine the points exceeding the preset distance error threshold, and give the corresponding missing area;

[0066] Integrate the distance errors of all point positions to determine the offset.

[0067] In a second aspect, the present invention further provides a detection device based on point cloud compression and cascaded rendering, which adopts the above-mentioned detection method based on point cloud compression and cascaded rendering. The device includes:

[0068] A data acquisition module, which is used to acquire point cloud data and a building information model of a construction site;

[0069] A scene determination module, which is used to determine a construction scene according to the building information model;

[0070] A data compression module, which is used to compress the point cloud data according to the construction scene and a pre-determined scene-compression matching strategy;

[0071] A data rendering module, which is used to render the compressed point cloud data according to the construction scene and a pre-determined scene-rendering matching strategy to generate a three-dimensional rendering;

[0072] A difference comparison module, which is used to compare the differences between the building information model and the three-dimensional rendering to obtain construction differences.

[0073] Further, the scene determination module includes:

[0074] Determine spatial information and process flow according to the building information model;

[0075] Traverse the equipment types involved in the process flow to determine whether there is a unique equipment corresponding to a certain construction scene;

[0076] When there is no unique equipment corresponding to a certain construction scene in the process flow, match the equipment type and spatial information with the scene characteristics of each construction scene, and give the corresponding matched construction scene.

[0077] Further, the scene determination module further includes:

[0078] Obtain the necessary equipment types and spatial limitation conditions of each construction scene;

[0079] Match the equipment type with the necessary equipment types of each construction scene, and give the initially matched construction scene;

[0080] Match the spatial information with the spatial limitation conditions corresponding to the initially matched construction scene, and give the construction scene corresponding to the spatial limitation conditions satisfied by the spatial information;

[0081] Based on the construction scene corresponding to the spatial limitation conditions satisfied by the spatial information, give the corresponding scoring rules and score the spatial information to determine the construction scene with the highest score.

[0082] Further, the detection device further includes a strategy determination module, which includes:

[0083] Obtain historical point cloud data corresponding to different construction scenarios and various point cloud compression algorithms;

[0084] Compress the historical point cloud data corresponding to each construction scenario respectively using various point cloud compression algorithms;

[0085] Calculate the data compression ratio and point cloud geometric deviation between the historical point cloud data compressed by different point cloud compression algorithms for all construction scenarios and the historical point cloud data before its own compression;

[0086] Determine the optimal point cloud compression algorithm corresponding to each construction scenario according to the data compression ratio and point cloud geometric deviation corresponding to the historical point cloud data compressed by different point cloud compression algorithms for each construction scenario;

[0087] Associate each construction scenario with the determined optimal point cloud compression algorithm.

[0088] Furthermore, the strategy determination module includes:

[0089] Comprehensively analyze the data compression ratio and point cloud geometric deviation to determine the compression scores corresponding to the historical point cloud data compressed by different point cloud compression algorithms for each construction scenario;

[0090] Select the point cloud compression algorithm corresponding to the highest compression score in each construction scenario.

[0091] Furthermore, the strategy determination module also includes:

[0092] Obtain historical point cloud data compressed by different construction scenarios through a pre-determined scenario-compression matching strategy and various cascaded rendering algorithms;

[0093] Render the historical point cloud data corresponding to each construction scenario respectively using various cascaded rendering algorithms to obtain historical 3D renderings;

[0094] Calculate the rendering scores for all historical 3D renderings and give the cascaded rendering algorithm with the highest rendering score corresponding to each construction scenario;

[0095] Associate each construction scenario with the cascaded rendering algorithm with the highest rendering score.

[0096] Furthermore, the strategy determination module includes:

[0097] Obtain the key equipment and corresponding spatial positions in each historical 3D rendering;

[0098] Determine the visual light and shadow difference degree based on each historical 3D rendering and the actual construction site image corresponding to the historical point cloud data;

[0099] Determine the detail difference degree based on each key device in each historical 3D rendering and the corresponding device before the historical point cloud data is compressed;

[0100] Determine the position deviation degree based on the spatial position of each key device in each historical 3D rendering and the spatial position before rendering;

[0101] Comprehensively analyze the visual light and shadow difference degree, detail difference degree, and position deviation degree of each historical 3D rendering respectively, and give the rendering score corresponding to each historical 3D rendering.

[0102] Furthermore, the strategy determination module includes:

[0103] Extract the lighting parameters and perform color conversion on the historical 3D renderings and the actual construction site images respectively, and calculate the brightness gradient change of the historical 3D renderings;

[0104] Compare the lighting parameters of the historical 3D renderings and the actual construction site images, and give the shadow matching degree;

[0105] Compare the color conversion results of the historical 3D renderings and the actual construction site images, and give the color difference;

[0106] Comprehensively analyze the color difference and the calculation result of the brightness gradient change after normalization with the shadow matching degree respectively, and give the visual light and shadow matching calculation result.

[0107] Furthermore, the strategy determination module includes:

[0108] Perform edge detection analysis and curvature change detection analysis on each historical 3D rendering respectively, and give the edge sharpness and surface fidelity;

[0109] Perform geometric shape matching analysis on each key device in each historical 3D rendering and the corresponding device before the historical point cloud data is compressed, and give the shape matching degree;

[0110] Comprehensively analyze the shape matching degree, edge sharpness, and surface fidelity, and give the detail difference result.

[0111] Furthermore, the difference comparison module includes:

[0112] Extract the features of the building information model and the 3D rendering respectively, and give their respective key feature points;

[0113] Match the key feature points of the building information model and the 3D rendering, and give the set of matched key feature point pairs;

[0114] Based on the set of matched key feature point pairs, determine the rigid transformation matrix and align the building information model and the 3D rendering;

[0115] Based on the rigid transformation matrix, perform point-by-point calculations on the aligned building information model and the 3D rendering diagram, and give the offset and missing areas.

[0116] Further, the difference comparison module includes:

[0117] Based on the rigid transformation matrix, determine the distance error of each corresponding point position in the building information model and the 3D rendering diagram;

[0118] Based on the distance error of each point position and the preset distance error threshold, determine the point positions that exceed the preset distance error threshold, and give the corresponding missing areas;

[0119] Integrate the distance errors of all point positions to determine the offset.

[0120] A detection method and device based on point cloud compression and cascaded rendering provided by the present invention have at least the following beneficial effects:

[0121] (1) Dynamically select the compression strategy based on the construction scene type to achieve the optimal solution of "accuracy - efficiency"; improve the rendering rate while ensuring visual integrity; that is, the present invention can effectively improve the efficiency of point cloud data processing and rendering in a complex construction environment, and realize the difference comparison between the construction site and the designed construction plan.

[0122] (2) By adopting a variety of point cloud compression algorithms and cascaded rendering technologies to construct a scene - compression matching strategy and a scene - rendering matching strategy, it is possible to intelligently select suitable compression and rendering schemes for different types of power construction scenes and quickly generate high - precision 3D rendering diagrams. Description of the Drawings

[0123] Figure 1 It is a flowchart of a detection method based on point cloud compression and cascaded rendering provided by the present invention;

[0124] Figure 2 It is a flowchart of pre - determining a scene - compression matching strategy provided by an embodiment of the present invention;

[0125] Figure 3 It is a flowchart of pre - determining a scene - rendering matching strategy provided by an embodiment of the present invention;

[0126] Figure 4 It is a flowchart of difference comparison provided by an embodiment of the present invention;

[0127] Figure 5 It is a schematic diagram of a detection device based on point cloud compression and cascaded rendering provided by the present invention. Detailed Embodiments

[0128] To better understand the above technical solutions, the following will provide a detailed description of the above technical solutions in conjunction with the accompanying drawings of the specification and specific embodiments. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative work fall within the scope of protection of the present invention.

[0129] The terms used in the embodiments of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The singular forms "a", "the" and "said" used in the embodiments of the present invention and the appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. "Plural" generally includes at least two.

[0130] It should also be noted that the term "comprising", "including" or any other variation thereof is intended to cover non-exclusive inclusion, so that a commodity or device including a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such commodity or device. Without further limitation, an element defined by the statement "including one..." does not exclude the existence of another identical element in the commodity or device including the said element.

[0131] With the continuous expansion of the construction scale of power transmission lines and substations, how to improve the monitoring accuracy and efficiency during the construction process has become a key issue. Point cloud technology has been widely applied in power engineering. However, the huge volume and complexity of point cloud data make it difficult for traditional processing and analysis methods to meet the requirements of efficient construction detection.

[0132] Currently, the construction comparison technology based on building information models and point cloud data is mainly used in related technologies. However, with the increase in project scale and complexity, the demand for real-time monitoring of construction progress and quality is growing. The comparison between traditional building information models and actual construction data mainly relies on manual measurement and static comparison, which can no longer meet the construction requirements.

[0133] In view of this, the present invention provides a construction plan comparison and detection method based on point cloud compression and cascaded rendering technology, which can effectively improve the efficiency of point cloud data processing and rendering in complex construction environments, and compare the differences between the construction site and the design plan. By adopting a variety of point cloud compression algorithms and cascaded rendering technology, this method can intelligently select suitable compression and rendering schemes for different types of power construction scenarios, quickly generate high-precision three-dimensional renderings, and accurately compare the construction site with the building information model through a difference detection algorithm. The present invention can not only significantly reduce the redundancy of point cloud data, improve data transmission and processing efficiency, but also provide visual feedback of dynamic construction scenarios through real-time rendering technology. Combining various functions such as equipment deviation detection and line laying monitoring, this method can help the construction team timely discover errors in the construction process, reduce rework and cost increases caused by non-conforming construction, and ultimately improve construction quality and efficiency.

[0134] As Figure 1 shown, the present invention provides a detection method based on point cloud compression and cascaded rendering, which may include:

[0135] Obtain the point cloud data of the construction site and the building information model of the corresponding construction plan;

[0136] Determine the construction scenario according to the building information model;

[0137] Compress the point cloud data according to the construction scenario and the pre-determined scenario-compression matching strategy;

[0138] Render the compressed point cloud data according to the construction scenario and the pre-determined scenario-rendering matching strategy to generate a three-dimensional rendering;

[0139] Perform a difference comparison between the building information model and the three-dimensional rendering to obtain construction differences, wherein the difference comparison is realized through a difference detection algorithm.

[0140] In the present invention, when obtaining the point cloud data of the construction site and the building information model of the corresponding construction plan, the point cloud data may include the equipment information and three-dimensional space information of the construction site.

[0141] It can be understood that obtaining the point cloud data and building information model of the construction site is the basis of the entire construction detection process. The point cloud data is obtained from the construction site through technical means such as laser scanning, and has high-precision three-dimensional space information and equipment layout data. The building information model (BIM), on the other hand, is a digital three-dimensional model of the construction plan, containing the design specifications, structural details and spatial arrangements of the project. By obtaining these data, the accuracy and consistency required in the subsequent comparison process can be ensured.

[0142] After obtaining the Building Information Model (BIM), the construction scenarios can be determined based on the BIM, including:

[0143] Determine the spatial information and technological processes according to the BIM;

[0144] Traverse the types of equipment involved in the technological processes to determine whether there is any equipment unique to a certain construction scenario;

[0145] When there is no equipment unique to a certain construction scenario in the technological processes, match the equipment types and spatial information with the scenario characteristics of each construction scenario, and give the corresponding matched construction scenarios, including: obtain the necessary equipment types and spatial limitation conditions of each construction scenario; match the equipment types with the necessary equipment types of each construction scenario to give the preliminarily matched construction scenarios; match the spatial information with the spatial limitation conditions corresponding to the preliminarily matched construction scenarios to give the construction scenarios corresponding to the spatial limitation conditions satisfied by the spatial information; based on the construction scenarios corresponding to the spatial limitation conditions satisfied by the spatial information, give the corresponding scoring rules and score the spatial information to determine the construction scenario with the highest score.

[0146] It can be understood that extracting the construction scenarios based on the obtained BIM is to clearly display the specific structures and equipment layouts in the construction site area. The acquisition of construction scenarios not only includes the layout of the physical space, but also includes information such as technological processes related to construction and the installation sequence of equipment. By combining the point cloud data with the BIM, a virtual scenario corresponding one-to-one to the construction site can be formed, providing a basis for subsequent algorithm selection.

[0147] When performing construction inspection, compare the spatial information in the BIM with the template characteristics of the construction scenarios. It is necessary to specifically extract spatial parameters and judge the scenario attribution through multi-level matching. In actual application scenarios, it includes:

[0148] The first step: Spatial information extraction

[0149] The spatial information in the BIM mainly includes parameters such as equipment positions, equipment spacings, spatial volumes, and structural characteristics. For example, in a certain construction scenario, the extracted spatial information is as follows: the main equipment coordinates in the point cloud data: Equipment A (transformer): (2.5m, 1.0m, 3.2m); Equipment B (high-voltage circuit breaker): (5.0m, 1.5m, 3.2m); Equipment C (busbar): (8.0m, 2.0m, 3.2m); Equipment spacing: The Euclidean distance between Equipment A and Equipment B is 2.7m, and the distance between Equipment B and Equipment C is 3.6m. Spatial dimensions: length 15m, width 10m, height 5m, total volume is 750m 3 . Wall thickness: 20cm, beam height is 50cm, and pipe inclination angle is 15°.

[0150] Step 2: Standardization of Template Features in Construction Scenarios

[0151] Different construction scenario templates store standard features, mainly including equipment types, equipment distribution rules, and spatial dimension ranges, determining the necessary equipment types and spatial constraint conditions they contain. The spatial constraint conditions include equipment distribution rules and spatial dimension ranges. For example, the template of the distribution network construction scenario: Necessary equipment types: Include transformers, high-voltage circuit breakers, and busbars. Equipment relative position requirements: The distance between the transformer and the high-voltage circuit breaker is 2.0m - 3.0m, and the distance between the busbar and the high-voltage circuit breaker is 3.0m - 4.0m. Spatial dimension range: 500m 3 ~1000m 3 , the wall thickness is 18cm - 22cm, and the beam height is 45cm - 55cm.

[0152] Step 3: Multi-level Matching Analysis

[0153] 1. Equipment-level Matching

[0154] Compare the equipment types and coordinates in the building information model one by one: If all necessary equipment types are successfully matched, it meets the template requirements. The distances between the equipment meet the standard requirements. The distance between the transformer and the high-voltage circuit breaker is 2.7m, and the distance between the busbar and the high-voltage circuit breaker is 3.6m, both within the specified range (±10%). Matching result: In multi-level matching, the weight of equipment-level matching is 0.4, and the final score of equipment-level matching is 40.

[0155] 2. Structural Feature Matching

[0156] The extracted wall thickness is 20cm, and the beam height is 50cm, both within the template standard range. Matching result: In multi-level matching, the weight of structural feature matching is 0.3, and the final score of structural feature matching is 28.

[0157] 3. Overall Space Matching

[0158] Calculate the difference in spatial volume. The template standard range of the construction scenario is 500m 3 ~1000m 3 , while the current spatial volume is 750m 3 , and the match is successful. The spatial layout is consistent with the template, and the functional area division is reasonable. Matching result: In multi-level matching, the weight of overall space matching is 0.3, and the final score of overall space matching is 27.

[0159] Step 4: Comprehensive Scoring and Judgment

[0160] According to the above matching analysis, the comprehensive scores are as follows: equipment matching: 40 / 40; structural feature matching: 28 / 30; overall space matching: 27 / 30; total score: 95 / 100; set the construction scene matching threshold (for example, 85), and since the comprehensive score exceeds 85 points, the construction scene is determined to be a distribution network construction scene. The matching results are accurate and can be used for subsequent point cloud compression and cascade rendering scene optimization. Among them, spatial information can be directly determined through point cloud data.

[0161] The present invention compresses point cloud data according to a construction scene and a predetermined scene-compression matching strategy, which may include: determining an optimal point cloud compression algorithm associated with the construction scene based on the predetermined scene-compression matching strategy; and compressing the point cloud data according to the determined optimal point cloud compression algorithm.

[0162] In an embodiment of the present invention, a point cloud compression algorithm that best matches the construction scene is selected through a scene-compression matching strategy, and the point cloud compression algorithm includes octree compression, compression based on inter-frame differences, and wavelet transform. It can be understood that based on the complexity and characteristics of different construction scenes, selecting the best point cloud compression algorithm can effectively reduce data redundancy and improve processing efficiency. The selection of the point cloud compression algorithm needs to be combined with the equipment distribution, regional scale, and real-time requirements of data processing at the construction site. For example, for the construction scene of a transmission line with long distances and sparse node distribution, octree compression can efficiently process spatial information in layers; while in the construction scene of a substation with dense equipment and complex details, wavelet transform helps to retain key high-frequency information. Compression based on inter-frame differences is suitable for scenes with more dynamic changes, such as the installation process of a wind power tower, which can highlight the dynamic changes in construction.

[0163] Compressing the point cloud data of the construction site based on the optimal point cloud compression algorithm can reduce the amount of data and improve the efficiency of subsequent rendering and processing. Different compression methods can retain key spatial and equipment information according to the needs of the construction scene. For example, octree compression is suitable for layered processing of large areas and can save detailed spatial information at different levels, while wavelet transform can retain high-frequency details of the structure at a more detailed level. In this compression step, the compressed point cloud data not only reduces data redundancy, but also maintains the accuracy of the original scene, laying the foundation for subsequent rendering and difference comparison.

[0164] like Figure 2 As shown, the pre-determination of the scene-compression matching strategy may include:

[0165] Obtain historical point cloud data corresponding to different construction scenarios and a variety of point cloud compression algorithms;

[0166] Compress the historical point cloud data corresponding to each construction scenario using a variety of point cloud compression algorithms respectively;

[0167] Calculate the data compression ratio and the point cloud geometric deviation between the historical point cloud data after compressing all construction scenarios using different point cloud compression algorithms and the historical point cloud data before its own compression;

[0168] Determine the best point cloud compression algorithm corresponding to each construction scenario according to the data compression ratio and the point cloud geometric deviation corresponding to the historical point cloud data after compressing each construction scenario using different point cloud compression algorithms;

[0169] Associate each construction scenario with the determined best point cloud compression algorithm.

[0170] Among them, the data compression ratio CR is determined by calculating the ratio of the data size before compression to the data size after compression. For example, CR = data size before compression / data size after compression; the point cloud geometric deviation PC satisfies the following relationship:

[0171]

[0172] Among them, and represent the point cloud positions before and after compression respectively, and N is the total number of point clouds.

[0173] Determine the best point cloud compression algorithm corresponding to each construction scenario according to the data compression ratio and the point cloud geometric deviation corresponding to the historical point cloud data after compressing each construction scenario using different point cloud compression algorithms, including: comprehensively analyzing the data compression ratio and the point cloud geometric deviation to determine the compression score corresponding to the historical point cloud data after compressing each construction scenario using different point cloud compression algorithms; selecting the point cloud compression algorithm corresponding to the highest compression score in each construction scenario; among them, the compression score satisfies the following relationship:

[0174]

[0175] In the formula, S is the compression score, α is the weight of the data compression ratio, β is the weight of the point cloud geometric deviation, CR is the data compression ratio, and PC is the point cloud geometric deviation.

[0176] After the present invention compresses point cloud data by selecting the optimal point cloud compression algorithm, it can render the compressed point cloud data according to the construction scenario and the pre-determined scenario-rendering matching strategy to generate a three-dimensional rendering. Specifically, it includes: determining a cascaded rendering algorithm associated with the construction scenario based on the scenario-rendering matching strategy; rendering the compressed point cloud data according to the determined cascaded rendering algorithm to generate a three-dimensional rendering. Among them, the scenario-rendering matching strategy is used to select the optimal cascaded rendering algorithm from multiple cascaded rendering algorithms according to the construction scenario. The cascaded rendering algorithms include hierarchical rendering with multiple levels of detail, spatial partitioning and block rendering, multi-view parallel rendering, and real-time rendering.

[0177] It can be understood that it is crucial to select the most suitable cascaded rendering algorithm according to the complexity, dynamics, and detail requirements of the construction scenario. Hierarchical rendering with multiple levels of detail can provide a distinction between near and far for the construction of long-distance transmission lines, combining high-resolution rendering in the near area with low-resolution rendering in the far area; spatial partitioning and block rendering is suitable for scenarios with dense equipment such as substations, which can render in regions and reduce the rendering burden; multi-view parallel rendering is suitable for dynamic scenarios, such as wind power construction, allowing the construction progress to be observed from multiple angles simultaneously; real-time rendering is suitable for construction scenarios that require immediate feedback, especially in photovoltaic power plants or maintenance construction, where a visual feedback of the construction status is needed quickly.

[0178] Based on the selected optimal cascaded rendering algorithm, the present invention renders the compressed point cloud data to generate a three-dimensional rendering with accurate space, equipment layout, and details. This three-dimensional rendering can display the actual state of the construction site and correspond to the design state in the building information model. By combining appropriate rendering algorithms, the complex structure, equipment installation location, and dynamic process of the construction site are accurately reflected in three-dimensional space, ensuring that subsequent comparisons can be based on high-precision image data.

[0179] Specifically, as Figure 3 shown, the pre-determination of the scenario-rendering matching strategy may include:

[0180] Obtaining the historical point cloud data compressed by different construction scenarios through the pre-determined scenario-compression matching strategy and multiple cascaded rendering algorithms;

[0181] Rendering the historical point cloud data compressed for each construction scenario respectively using multiple cascaded rendering algorithms to obtain historical three-dimensional renderings;

[0182] Calculating the rendering scores for all historical three-dimensional renderings and giving the cascaded rendering algorithm with the highest rendering score for each construction scenario;

[0183] Associating each construction scenario with the cascaded rendering algorithm with the highest rendering score.

[0184] Among them, calculating the rendering score for all historical 3D rendering diagrams may include:

[0185] Obtain the key devices and corresponding spatial positions in each historical 3D rendering diagram;

[0186] Determine the visual light and shadow difference degree based on each historical 3D rendering diagram and the actual construction site image corresponding to the historical point cloud data;

[0187] Determine the detail difference degree based on each key device in each historical 3D rendering diagram and the corresponding device before the compression of the historical point cloud data; among them, the key device can be all complete devices included in the historical 3D rendering diagram, or it can also be some devices with specific structures;

[0188] Determine the position deviation degree based on the spatial position of each key device in each historical 3D rendering diagram and the spatial position before rendering;

[0189] Comprehensively analyze the visual light and shadow difference degree, detail difference degree, and position deviation degree of each historical 3D rendering diagram respectively, and give the rendering score corresponding to each historical 3D rendering diagram; among them, the comprehensive analysis is to calculate the final rendering score by superimposing the corresponding weights of the determined visual light and shadow difference degree, detail difference degree, and position deviation degree.

[0190] Specifically, determining the visual light and shadow difference degree based on each historical 3D rendering diagram and the actual construction site image corresponding to the historical point cloud data may include:

[0191] Extract the lighting parameters and perform color conversion on the historical 3D rendering diagram and the actual construction site image respectively, and calculate the brightness gradient change of the historical 3D rendering diagram;

[0192] Compare the lighting parameters of the historical 3D rendering diagram and the actual construction site image, and give the shadow matching degree; specifically, the shadow area of the historical 3D rendering diagram can be identified by the brightness threshold method, and then the shadow area of the historical 3D rendering diagram is overlapped and matched with the shadow area corresponding to the compressed point cloud data to give the shadow matching degree SA = SA0 / (SA1 + SA2 - SA0), where SA0 is the overlapping area of the shadow area of the historical 3D rendering diagram and the shadow area corresponding to the compressed point cloud data, SA1 is the shadow area of the historical 3D rendering diagram, and SA2 is the shadow area corresponding to the compressed point cloud data;

[0193] Compare the color conversion results of the historical 3D rendering diagram and the actual construction site image, and give the color difference; specifically, the historical 3D rendering diagram and the actual construction site image can be converted to the Lab space, and the mean difference is calculated; the standard deviation of the hue distribution is statistically analyzed to determine the color deviation degree (color difference), which satisfies the following relationship:

[0194]

[0195] Wherein, ΔE Lab is the degree of color deviation, i.e., the color difference, and I render is the historical three-dimensional rendering, and L render , a render , b render are respectively the lightness, hue and saturation of the historical three-dimensional rendering in the Lab color space, and I real is the image of the actual construction site, and L real , a real , b real are respectively the lightness, hue and saturation of the image of the actual construction site in the Lab color space;

[0196] After normalizing the calculation results of the color difference and the brightness gradient change respectively, a comprehensive analysis is carried out with the shadow matching degree to give the calculation result of the visual light and shadow matching; among them, the comprehensive analysis is obtained by superimposing and calculating through the corresponding weights of the determined color difference, brightness gradient change and shadow matching degree to obtain the final visual light and shadow matching score.

[0197] Calculating the brightness gradient change of the historical three-dimensional rendering may include:

[0198] Converting the historical three-dimensional rendering into a grayscale image, satisfying the following relationship:

[0199] L(x,y) = α L1 R(x,y) + α L2 G(x,y) + α L3 B(x,y)

[0200] Wherein, L(x,y) is the grayscale value of the grayscale image at the x,y coordinate point, and α L1 , α L2 , α L3 are respectively the weight values of the red, green and blue color channels, which can be 0.299, 0.587, 0.114 respectively, and R(x,y), G(x,y), B(x,y) are respectively the values of the red, green and blue colors in the historical three-dimensional rendering;

[0201] Using the Sobel operator to extract the gradient amplitude of the grayscale image, satisfying the following relationship:

[0202]

[0203] Wherein, GT(x,y) is the gradient amplitude, and GT x and GT y are respectively the horizontal gradient and the vertical gradient, and CK x and CK yThey are the horizontal gradient kernel and the vertical gradient kernel, CK x and CK y can be preset. For example:

[0204]

[0205] Divide the gradient magnitude into several intervals and give the pixel proportion in each interval;

[0206] Based on the pixel proportions in all intervals, give the luminance entropy, which satisfies the following relationship:

[0207]

[0208] In the formula, BE is the luminance entropy, B is the number of divided intervals, and h(k) is the pixel proportion of the k-th interval;

[0209] Combine the luminance entropy and the divided intervals to determine the luminance gradient change.

[0210] Among them, the number of divided intervals can be adjusted according to specific requirements, and the influence on the scoring result can be verified through experiments. Preferably, the gradient magnitude is divided into 32 intervals. In addition, when normalizing the calculation result of the luminance gradient change, the pixel proportion in each interval can be normalized first; calculate the luminance entropy for the normalized pixel proportion; determine the luminance gradient change score based on the luminance entropy calculation result and the divided intervals, which satisfies the following relationship: LC = 100 * BE / log2B, where LC is the luminance gradient change score.

[0211] The normalization of the color difference can be determined by the ratio of the color difference to the color difference range.

[0212] When the present invention determines the detail differences based on each key device in each historical three-dimensional rendering diagram and the corresponding device before the historical point cloud data is compressed, it may include:

[0213] Perform edge detection analysis and curvature change detection analysis on each historical three-dimensional rendering diagram respectively, and give the edge sharpness and surface fidelity;

[0214] Perform geometric shape matching analysis on each key device in each historical three-dimensional rendering diagram and the corresponding device before the historical point cloud data is compressed, and give the shape matching degree;

[0215] Perform comprehensive analysis on the shape matching degree, edge sharpness and surface fidelity, and give the detail difference result; among them, the comprehensive analysis is to calculate the weighted sum of the determined shape matching degree, edge sharpness and surface fidelity to obtain the final detail difference score.

[0216] Among them, when performing edge detection analysis, the Sobel operator and / or the Canny edge detection algorithm can be used to calculate the total edge length and clarity in the historical three-dimensional rendering, so as to achieve the smoothness and sharpness of the detected edges. Taking the calculation using the Sobel operator and the Canny edge detection algorithm as an example, it specifically includes: converting the historical three-dimensional rendering into a grayscale image, and calculating the gradient magnitude through the Sobel operator on the grayscale image (this step can adopt the above grayscale image conversion and gradient magnitude calculation process); performing Gaussian filtering on the grayscale image through the Canny edge detection algorithm to smooth the image and calculating the gradient magnitude after non-maximum suppression; performing hysteresis processing on the gradient magnitude after non-maximum suppression according to the preset strong edge threshold and weak edge threshold to generate a binary edge map; performing contour extraction on the binary edge map to give a set of continuous edge segments; determining the edge clarity based on the gradient magnitude and the edge segment set, satisfying the following relationship:

[0217]

[0218] In the formula, S E is the edge clarity, L total is the total edge length, GT x and GT y are the horizontal gradient and vertical gradient calculated by the Sobel operator respectively, H E , W E are the height and width of the image in the Sobel operator calculation process respectively, E is the binary edge map processed by the Canny edge detection algorithm, (x E,i , y E,i ) is the edge pixel coordinate of the i-th point in the binary edge map.

[0219] When performing surface fidelity analysis, surface normal vector analysis can be used to detect the curvature change of the surface in the historical three-dimensional rendering, and calculate the smoothness and fidelity of the details, specifically including: predicting the normal vector field from the historical three-dimensional rendering using methods such as SFTNet or DeepSS; giving the Gaussian curvature and the corresponding curvature variance based on the normal vector field; giving the surface fidelity based on the curvature variance, satisfying the following relationship:

[0220]

[0221] In the formula, F K is the surface fidelity, is the curvature variance, is the true surface curvature variance, which can be calculated in advance or obtained through a reference dataset, is the curvature mean, K(x, y) is the Gaussian curvature, (Nx, Ny, Nz) are the coordinates of the normal vector field, K ijIt is the curvature tensor component of the historical 3D rendering diagram, determined by calculating the second derivative. i, j = [x, y, z].

[0222] When performing shape matching degree analysis, calculate the Hausdorff distance between the historical 3D rendering diagram and the device geometric form in the original point cloud data (compressed point cloud data) to evaluate the matching degree of the shape, and the score range is from 0 to 1. Specifically, it includes: extracting the 3D point cloud data of the key device from the historical 3D rendering diagram; calculating the distance between the extracted 3D point cloud data of the key device and the corresponding compressed point cloud data of the key device; determining the shape matching degree according to the calculated distance and the maximum distance, and satisfying the following relationship:

[0223]

[0224] In the formula, G D is the shape matching degree, D max is the maximum distance, which can be determined according to the point cloud range, such as the diagonal length of the point cloud bounding box, D H (P render , P real ) is the distance between the extracted 3D point cloud data of the key device and the corresponding compressed point cloud data of the key device, P render is the set of the extracted 3D point cloud data of the key device, P real is the set of the corresponding compressed point cloud data of the key device, p is the extracted 3D point cloud data of the key device, and q is the corresponding compressed point cloud data of the key device.

[0225] Among them, the construction scenarios include transmission line construction, substation construction, distribution network construction, wind power generation construction, photovoltaic power generation construction, and power maintenance and transformation construction. It can be understood that these construction scenarios involve different equipment layouts, spatial scales, and dynamic requirements, so different point cloud compression algorithms can be used to optimize the data processing efficiency and accuracy.

[0226] In actual application scenarios, the correspondence between construction scenarios and the optimal point cloud compression algorithms in the scenario-compression matching strategy includes: when the construction scenario is transmission line construction, the octree compression method is selected; when the construction scenario is substation construction or power maintenance and transformation construction, wavelet transform is selected; when the construction scenario is wind power generation construction, the compression method based on inter-frame difference is selected; when the construction scenario is distribution network or photovoltaic power generation construction scenario, octree compression combined with wavelet transform is selected. It can be understood that when the construction scenario is transmission line construction, the transmission line usually extends relatively long and has a relatively simple structure, but has a wide coverage area. Therefore, the octree compression method is selected. This method can efficiently process a large amount of spatial information in layers, adapt to the characteristics of long-distance linear distribution, and can perform hierarchical processing on different regions at different distances, thereby optimizing data storage and transmission efficiency. The correspondence between construction scenarios and the optimal cascaded rendering algorithms in the scenario-rendering matching strategy includes: when the construction scenario is transmission line construction, the multi-level of detail hierarchical rendering method is selected; when the construction scenario is substation construction, the spatial partitioning and block rendering method is adopted; when the construction scenario is wind power generation construction, the multi-view parallel rendering method is selected; when the construction scenario is photovoltaic power generation construction scenario, the real-time rendering method is selected; when the construction scenario is distribution network construction or power maintenance and transformation construction, the spatial partitioning and block rendering combined with multi-view rendering method is used for parallel rendering.

[0227] Specifically, in view of the characteristics of transmission line construction, the present invention proposes an optimized scheme for the multi-level of detail hierarchical rendering technology, which is particularly suitable for the real-time visualization of long-distance construction scenarios. In long-distance transmission line construction, the equipment and construction operations in the near distance require high-precision rendering, while the scenes in the far distance can use low-precision simplified models to reduce the rendering calculation burden. The construction scenario is divided into three rendering levels: near, middle, and far: Near-distance area: High-precision rendering is used, including equipment details, personnel actions, etc. Middle-distance area: Medium-precision rendering is adopted, only showing key facilities and structures. Far-distance area: Simplified models and textures are used to ensure the display of the overall layout of the scene but reduce details. As the construction perspective changes, the rendering level of each area is automatically adjusted. When the user's perspective approaches the far-distance area, high-precision models and textures will be automatically loaded, and the precision of the rear scene will be gradually reduced. By performing spatial partitioning and block rendering on the entire construction scenario, the smoothness of rendering can be maintained while processing ultra-long-distance transmission lines. Each block area is independently rendered according to the perspective needs, avoiding performance bottlenecks caused by overall rendering.

[0228] When the construction scenario is substation construction or power maintenance and renovation construction, these scenarios usually have complex, dense equipment and diverse structures, and more detailed information needs to be retained. Therefore, wavelet transform is selected. Wavelet transform can retain detailed information while compressing, and is particularly suitable for complex three-dimensional equipment scenarios to ensure that the compressed point cloud data can completely present key details such as equipment installation and connection. When the construction scenario is wind power construction, the installation and commissioning of wind turbines have the characteristics of dynamic changes, especially the rotation of the blades and the change of the tower height. Therefore, a compression method based on inter-frame difference is selected. This method can effectively capture the changes at different time points during the construction process and is suitable for processing scenarios with changes in wind turbine tower height or equipment movement.

[0229] Specifically, in the wind power generation scenario, the position and attitude of the construction object (such as the wind turbine blade) are constantly changing. To effectively compress and save the dynamic data of these changes, the present invention proposes a compression method based on inter-frame difference. First, two adjacent frames of point cloud data are compared to calculate the displacement, attitude change, and possible structural change of each point. For the stationary or less-changing parts, they are marked as low priority, and high-ratio compression or omission of unnecessary data is used during compression. For the significantly-changing parts (such as the installation of the blade), they are preferentially recorded and saved at a lower compression rate. During the compression process, the key frames during the construction process are automatically identified, such as the start and end of the installation of the wind turbine blade. These key frames retain more data information to ensure that the key steps of the construction can be completely saved. According to the changes in the construction stage and scenario, the compression rate can be dynamically adjusted. During the rapid installation or movement process, the compression rate is reduced to save more details; during the stationary stage, the compression rate is increased to save storage space.

[0230] When the construction scenario is a distribution network or photovoltaic power generation construction scenario, these scenarios involve both large-scale spatial layouts and the need to retain certain detailed information. Therefore, octree compression combined with wavelet transform is selected. Octree compression can efficiently process the overall distribution of large-scale distribution networks and photovoltaic panels, while wavelet transform can retain the detailed information of local equipment installation to ensure the integrity and accuracy of the data.

[0231] In this embodiment, according to the real-time changes of the construction scenario, the best cascaded rendering algorithm is dynamically adjusted and reselected among multiple cascaded rendering algorithms. This process ensures that the rendering process can timely switch to the most suitable rendering method according to the changes in the construction environment, especially the construction progress or equipment movement. For example, during the installation of a wind power tower, when the installation stage of the wind turbine blade begins, the rendering method may switch from static multi-level-of-detail rendering to multi-view parallel rendering to better display the changes in dynamic construction.

[0232] Such as Figure 4As shown in the figure, after obtaining the 3D rendering, the present invention can perform a difference comparison between the building information model and the 3D rendering to obtain construction differences, specifically including:

[0233] Extract features from the building information model and the 3D rendering respectively, and give their respective key feature points;

[0234] Match the key feature points of the building information model and the 3D rendering, and give the set of matched key feature point pairs;

[0235] In the actual application scenario, performing a difference comparison between the building information model and the 3D rendering to obtain construction differences may include:

[0236] Extract features from the building information model and the 3D rendering respectively through the Harris3D algorithm or the ISS (Intrinsic Shape Signature) algorithm to obtain the corresponding set of high-curvature feature points; through this step, the points with high-curvature changes (i.e., key feature points) in the building information model and the 3D rendering can be identified, such as the corners of walls, the edges of doors and windows, etc., to complete corner detection;

[0237] Use the RANSAC (Random Sample Consensus) method to fit the plane equation for the building information model and the 3D rendering, extract the plane intersection line, and use it as an auxiliary feature to identify the plane features in the building (such as walls, floors, ceilings), to complete plane feature extraction. After completing the plane feature extraction, it can provide a matching benchmark for the calculation of the rigid transformation matrix between the 3D rendering and the building information model;

[0238] For each high-curvature feature point in the high-curvature feature point set corresponding to the Building Information Modeling (BIM) and the 3D rendering, calculate the descriptor. Specifically, FPFH (Fast Point Feature Histograms) and SHOT (Signature of Histograms of Orientations) can be used to calculate the descriptor. Among them, FPFH (Fast Point Feature Histograms) forms a feature histogram by calculating the local geometric structure descriptor of each high-curvature feature point for matching; SHOT (Signature of Histograms of Orientations) generates a descriptor through the normal vector distribution of the local geometric shape and has good matching effect, which is very suitable for the building scene. In the actual application scenario, when using FPFH (Fast Point Feature Histograms) to calculate the descriptor, it can include: constructing a KD-Tree to search the neighborhood of each point; calculating the normal vector histogram of the neighborhood points; normalizing the histogram to generate a feature vector. The generated feature vector is the descriptor. When using SHOT (Signature of Histograms of Orientations) to calculate the descriptor, it can include: calculating the normal vector direction histogram of the point neighborhood; performing Gaussian kernel smoothing on the histogram to generate rotation-invariant features. The generated rotation-invariant features are the descriptors.

[0239] Match the high-curvature feature points in the high-curvature feature point set corresponding to the Building Information Modeling (BIM) and the 3D rendering through the descriptors to determine the corresponding point pairs in the two high-curvature feature point sets and obtain the key feature point pair set. Specifically, by calculating the Euclidean distance between the descriptors corresponding to the high-curvature feature points, the point pair with the smallest distance is regarded as the preliminary matching point pair. Remove the incorrect preliminary matching point pairs. Specifically, the K-D tree can be used to quickly search for the nearest neighbor points, and the high-confidence matching point pairs can be screened out by the Nearest Neighbor Distance Ratio (NNDR) method.

[0240] Based on the matching key feature point pair set, determine the rigid transformation matrix and align the Building Information Modeling (BIM) and the 3D rendering.

[0241] The ICP (Iterative Closest Point) algorithm can be used to calculate the optimal rigid transformation matrix T to achieve the precise alignment of the Building Information Modeling (BIM) and the 3D rendering. The ICP algorithm can iteratively adjust the position and attitude of the data of the Building Information Modeling (BIM) and the compressed point cloud data, so that the two are in the same coordinate system and reference point, thereby eliminating the offset problem caused by the coordinate system difference. This step is crucial for subsequent difference detection to ensure the consistency of the data in the comparison process. Specific steps:

[0242] Initial transformation estimation: Using the set of matched point pairs, calculate the rotation matrix R and translation vector t of the initial rigid transformation matrix through SVD (Singular Value Decomposition);

[0243]

[0244] where T is the initial rigid transformation matrix, and P s,i is the corresponding point of the i-th matched 3D rendering in the set of matched point pairs, and P t,i is the corresponding point of the i-th matched Building Information Modeling (BIM) in the set of matched point pairs. n is the number of matched point pairs. and are the centroids of the corresponding points of the 3D rendering and Building Information Modeling in the matched point pairs respectively;

[0245] Iterative optimization: Based on nearest neighbor search, continuously iterate and update the point pairs, minimize the Euclidean distance error e between the two sets of points, use quaternion differentiation to correct the rotation R, and optimize the translation t based on the corrected rotation R to determine the optimal rigid transformation matrix.

[0246] Based on the rigid transformation matrix, perform point-by-point calculations on the aligned Building Information Modeling and 3D rendering to give the offset and missing areas, including:

[0247] Based on the determined optimal rigid transformation matrix, determine the distance error of each corresponding point position in the Building Information Modeling and 3D rendering; based on the distance error of each point position and the preset distance error threshold, determine the point positions that exceed the preset distance error threshold, and give the corresponding missing areas; integrate the distance errors of all point positions to determine the offset.

[0248] It can be understood that in this embodiment, the Building Information Modeling (BIM) is compared with the generated 3D rendering to detect the differences from the design scheme during the construction process. Deviations in position, angle, size, etc. between the construction site and the design model can be identified, such as problems where equipment installation does not conform to the design or there are deviations in the laying position of lines. In this way, any anomalies during the construction process can be timely identified and visually displayed, providing clear guidance for construction managers, facilitating timely correction of construction deviations, and thus improving construction quality and consistency.

[0249] In the embodiment of the present invention, after the comparison, it is also possible to generate a 3D heat map to display the construction differences. The 3D heat map includes high-difference areas and low-difference areas.

[0250] It can be understood that the results of difference detection can be presented in the form of a three-dimensional heat map to more intuitively display construction differences. The three-dimensional heat map visualizes the difference areas at the construction site according to the magnitude of the detected deviations. High-difference areas, such as installation errors or severely deviated equipment positions, will be shown in prominent colors, while low-difference areas, such as parts that conform to the design, will be identified with different colors. This visualization method can help construction management personnel quickly identify and focus on the problem areas, thereby optimizing the construction correction and adjustment processes.

[0251] Specifically, during the process of difference detection, a three-dimensional heat map will be generated according to the difference comparison results. The three-dimensional heat map can intuitively display construction differences and help construction personnel quickly identify problem areas. First, difference comparison is carried out to identify the difference areas. For each key node, errors in multiple dimensions such as its position, angle, and size are calculated, and the error data is converted into heat map colors. Larger difference points are shown in red, and the areas with high height deviations are marked as high-difference areas; smaller difference points are marked as low-difference areas and shown in green. The heat map can not only display position deviations but also various differences such as dimensions, angles, and equipment installation errors. This enables construction personnel to understand the problems from multiple perspectives and optimize the construction adjustment strategy. A threshold is automatically set according to the magnitude of the deviation data to distinguish deviations of different severity levels. As the construction progresses, the three-dimensional heat map can be dynamically updated to display the latest construction differences at the current stage. This helps the construction team understand the current construction situation at any time and quickly make corrections.

[0252] As Figure 5 shown, the present invention also provides a detection device based on point cloud compression and cascaded rendering, adopting the above-mentioned detection method based on point cloud compression and cascaded rendering. The device includes:

[0253] A data acquisition module, which is used to acquire the point cloud data of the construction site and the building information model corresponding to the construction plan;

[0254] A scene determination module, which is used to determine the construction scene according to the building information model;

[0255] A data compression module, which is used to compress the point cloud data according to the construction scene and the pre-determined scene-compression matching strategy;

[0256] A data rendering module, which is used to render the compressed point cloud data according to the construction scene and the pre-determined scene-rendering matching strategy to generate a three-dimensional rendering;

[0257] A difference comparison module, which is used to compare the building information model and the three-dimensional rendering to obtain construction differences.

[0258] Although the preferred embodiments of the present invention have been described, additional changes and modifications can be made to these embodiments by those skilled in the art once they learn the basic creative concept. Therefore, the appended claims are intended to be construed to include the preferred embodiments as well as all changes and modifications falling within the scope of the present invention. Obviously, those skilled in the art can make various changes and variations to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalent technologies, the present invention is also intended to include these modifications and variations.

Claims

1. A detection method based on point cloud compression and cascaded rendering, characterized in that, Including: Obtain the point cloud data of the construction site and the Building Information Modeling (BIM); Determine the construction scenario according to the BIM; Compress the point cloud data according to the construction scenario and the pre-determined scenario-compression matching strategy; Render the compressed point cloud data according to the construction scenario and the pre-determined scenario-rendering matching strategy to generate a three-dimensional rendering; Compare the differences between the BIM and the three-dimensional rendering to obtain the construction differences.

2. The detection method based on point cloud compression and cascaded rendering according to claim 1, characterized in that, Determine the construction scenario according to the BIM, including: Determine the spatial information and the technological process according to the BIM; Traverse the equipment types involved in the technological process to determine whether there is any equipment unique to a certain construction scenario; When there is no equipment unique to a certain construction scenario in the technological process, match the equipment type and the spatial information with the scenario characteristics of each construction scenario, and give the corresponding matched construction scenario.

3. The detection method based on point cloud compression and cascaded rendering according to claim 2, characterized in that, Match the equipment type and the spatial information with the scenario characteristics of each construction scenario, and give the corresponding matched construction scenario, including: Obtain the necessary equipment type and the spatial limitation conditions of each construction scenario; Match the equipment type with the necessary equipment type of each construction scenario, and give the preliminarily matched construction scenario; Match the spatial information with the spatial limitation conditions corresponding to the preliminarily matched construction scenario, and give the construction scenario corresponding to the spatial limitation conditions satisfied by the spatial information; Based on the construction scenario corresponding to the spatial limitation conditions satisfied by the spatial information, give the corresponding scoring rules and score the spatial information to determine the construction scenario with the highest score.

4. The detection method based on point cloud compression and cascaded rendering according to claim 1, wherein The pre-determination of the scenario-compression matching strategy includes: Obtain the historical point cloud data corresponding to different construction scenarios and various point cloud compression algorithms; Compress the historical point cloud data corresponding to each construction scenario respectively using various point cloud compression algorithms; Calculate the data compression ratio and the point cloud geometric deviation between the historical point cloud data compressed by different point cloud compression algorithms for all construction scenarios and the historical point cloud data before compression; Determine the best point cloud compression algorithm corresponding to each construction scenario according to the data compression ratio and the point cloud geometric deviation of the historical point cloud data compressed by different point cloud compression algorithms for each construction scenario; Associate each construction scenario with the determined best point cloud compression algorithm.

5. The detection method based on point cloud compression and cascaded rendering according to claim 1, wherein The pre-determination of the scenario-rendering matching strategy includes: Obtain the historical point cloud data compressed by the pre-determined scenario-compression matching strategy for different construction scenarios and various cascaded rendering algorithms; Render the historical point cloud data corresponding to each construction scenario respectively using various cascaded rendering algorithms to obtain historical three-dimensional renderings; Calculate the rendering scores for all historical three-dimensional renderings, and give the cascaded rendering algorithm with the highest rendering score corresponding to each construction scenario; Associate each construction scenario with the cascaded rendering algorithm with the highest rendering score.

6. The detection method based on point cloud compression and cascaded rendering according to claim 5, wherein Calculate the rendering scores for all historical three-dimensional renderings, including: Obtain the key equipment and the corresponding spatial positions in each historical three-dimensional rendering; Determine the visual light and shadow difference degree based on each historical three-dimensional rendering and the actual construction site image corresponding to the historical point cloud data; Determine the detail difference degree based on each key device in each historical 3D rendering and the corresponding device before the historical point cloud data is compressed; Determine the position deviation degree based on the spatial position of each key device in each historical 3D rendering and the spatial position before rendering; Comprehensively analyze the visual light and shadow difference degree, detail difference degree, and position deviation degree of each historical 3D rendering respectively, and give the rendering score corresponding to each historical 3D rendering.

7. The detection method based on point cloud compression and cascaded rendering according to claim 6, wherein, Determine the visual light and shadow difference degree based on each historical 3D rendering and the actual construction site image corresponding to the historical point cloud data, including: Extract the light parameters and perform color conversion on the historical 3D rendering and the actual construction site image respectively, and calculate the brightness gradient change of the historical 3D rendering; Compare the light parameters of the historical 3D rendering and the actual construction site image, and give the shadow matching degree; Compare the color conversion results of the historical 3D rendering and the actual construction site image, and give the color difference; Comprehensively analyze the color difference and the calculation result of the brightness gradient change after normalization with the shadow matching degree respectively, and give the visual light and shadow matching calculation result.

8. The detection method based on point cloud compression and cascaded rendering according to claim 6, wherein Determine the detail difference based on each key device in each historical 3D rendering and the corresponding device before the historical point cloud data is compressed, including: Perform edge detection analysis and curvature change detection analysis on each historical 3D rendering respectively, and give the edge sharpness and surface fidelity; Perform geometric shape matching analysis on each key device in each historical 3D rendering and the corresponding device before the historical point cloud data is compressed, and give the shape matching degree; Comprehensively analyze the shape matching degree, edge sharpness, and surface fidelity, and give the detail difference result.

9. The detection method based on point cloud compression and cascaded rendering according to any one of claims 1-8, characterized in that, Compare the differences between the building information model and the 3D rendering to obtain the construction differences, including: Extract features from the building information model and the 3D rendering respectively, and give their respective key feature points; Match the key feature points of the building information model and the 3D rendering, and give the set of matched key feature point pairs; Based on the set of matched key feature point pairs, determine the rigid transformation matrix, and align the building information model and the 3D rendering; Based on the rigid transformation matrix, perform point-by-point calculation on the aligned building information model and 3D rendering, and give the offset and missing area.

10. A detection device based on point cloud compression and cascaded rendering, characterized in that, Adopt the detection method based on point cloud compression and cascaded rendering as described in any one of claims 1-9, and the device includes: A data acquisition module for acquiring the point cloud data of the construction site and the building information model; A scene determination module for determining the construction scene according to the building information model; A data compression module for compressing the point cloud data according to the construction scene and the pre-determined scene-compression matching strategy; A data rendering module for rendering the compressed point cloud data according to the construction scene and the pre-determined scene-rendering matching strategy to generate a 3D rendering; A difference comparison module for comparing the differences between the building information model and the 3D rendering to obtain the construction differences.

Images