A method and system for controlling real scene modeling in a cloud platform

Abstract

The present invention provides a method and system for controlling real-scene modeling in a cloud platform, including: collecting real-scene data based on a preset shooting mode by using a drone and uploading the real-scene data to the cloud platform; extracting the composition features of the real-scene data based on the cloud platform, and splitting the real-scene data based on the composition features to obtain a set of modeling tasks; distributing the set of modeling tasks to each modeling node in a distributed manner, and performing dynamic interaction analysis on each modeling task based on the modeling node, and coupling and coordinating the analysis results to construct a real-scene model; performing multi-terminal visualization configuration on the real-scene model to complete real-scene modeling. The efficiency of real-scene modeling is improved, the accurate and effective construction of the real-scene model is realized, it is ensured that the real-scene model can operate effectively on multiple different terminals, and the effect and accuracy of real-scene modeling are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and particularly to a cloud platform real scene modeling control method and system. Background Art

[0002] With the development of information technology, the demand for real scene modeling in many fields is increasing day by day, such as urban planning, geographic information, architectural design, etc.;

[0003] In traditional real scene modeling, there may be problems such as unreasonable allocation of computing resources, low data processing efficiency, slow model construction speed, etc. At the same time, the needs and requirements of different users and projects for real scene modeling are also different. Moreover, when collecting real scene data, the collection of real scene data is not comprehensive and accurate enough, thus greatly reducing the effect and accuracy of real scene modeling;

[0004] Therefore, in order to overcome the above defects, the present invention provides a cloud platform real scene modeling control method and system. Summary of the Invention

[0005] The present invention provides a cloud platform real scene modeling control method and system, which is used to control an unmanned aerial vehicle to collect real scene data according to a preset shooting mode, so as to comprehensively and effectively obtain real scene data, and upload the collected real scene data to the cloud platform, facilitating the cloud platform to process the real scene data and providing reliable data support for real scene modeling. Secondly, the real scene data is split by the cloud platform to accurately and effectively determine a set of modeling tasks, and the modeling tasks are distributed to each modeling node in a distributed manner, facilitating each modeling node to synchronously execute the modeling tasks, improving the efficiency of real scene modeling. When performing real scene modeling, dynamic interaction and analysis are carried out between each modeling node to ensure the accuracy and reliability of the models finally obtained by each modeling node. Finally, the results obtained by each modeling node are coupled and coordinated to accurately and effectively construct a real scene model, and multi-terminal visualization configuration is performed on the real scene model to ensure that the real scene model can operate effectively on multiple different terminals, improving the effect and accuracy of real scene modeling.

[0006] The present invention provides a cloud platform real scene modeling control method, including:

[0007] Step 1: Based on the unmanned aerial vehicle, collect real scene data according to a preset shooting mode, and upload the real scene data to the cloud platform;

[0008] Step 2: Based on the cloud platform, extract the composition features of the real scene data, and split the real scene data based on the composition features to obtain a set of modeling tasks;

[0009] Step 3: Distribute the modeling task set to each modeling node in a distributed manner, perform dynamic interaction analysis on each modeling task based on the modeling node, and couple and coordinate the analysis results to construct a real-scene model;

[0010] Step 4: Perform multi-terminal visualization configuration on the real-scene model to complete real-scene modeling.

[0011] Preferably, for a cloud platform real-scene modeling control method, in step 1, collecting real-scene data based on a drone according to a preset shooting mode includes:

[0012] Obtain the requirements for real-scene data collection, and analyze the requirements for real-scene data collection to obtain the collection transformation angle and collection accuracy of the tilt camera on the drone;

[0013] At the same time, obtain the scene distribution characteristics of the area to be collected, and determine the flight route of the drone over the area to be collected based on the scene distribution characteristics and the collection accuracy;

[0014] Based on the collection change angle and the flight route, obtain the preset shooting mode of the drone, and control the drone to collect real-scene data based on the preset shooting mode.

[0015] Preferably, for a cloud platform real-scene modeling control method, controlling the drone to collect real-scene data based on a preset shooting mode includes:

[0016] Obtain the collected real-scene data, and perform edge contour detection on the real-scene images of the real-scene data to obtain the edge contour bounding rectangle of the real-scene image at each collection angle;

[0017] Based on the flight route of the drone over the area to be collected, determine adjacent image groups of the real-scene images of the real-scene data, and perform opposite overlap on the real-scene images in the adjacent image groups;

[0018] Based on the opposite overlap, determine the overlap degree of the edge contour bounding rectangles of adjacent real-scene images, and when the overlap degree does not meet the preset threshold, re-collect the real-scene data;

[0019] Otherwise, retain the real-scene images in the adjacent image groups, and add position labels to the real-scene images based on the retention results.

[0020] Preferably, for a cloud platform real-scene modeling control method, in step 1, uploading the real-scene data to the cloud platform includes:

[0021] Obtain the communication parameters of the local device, and dock the local device and the drone based on the communication parameters;

[0022] Based on the docking result, cache the real-scene data collected by the drone on the local device, and construct a communication link between the local device and the cloud platform based on the caching result;

[0023] Compress the cached real - scene data, and split the compressed real - scene data to obtain compressed data blocks;

[0024] Upload the compressed data blocks to the cloud platform in sequence based on the communication link.

[0025] Preferably, for a cloud - platform real - scene modeling control method, in step 2, extract the compositional features of the real - scene data based on the cloud platform, and split the real - scene data based on the compositional features to obtain a set of modeling tasks, including:

[0026] Obtain the obtained real - scene data, and analyze the real - scene data based on the scene modeling knowledge system to obtain the semantic information of each single - entity data in the real - scene data;

[0027] Determine the spatial position relationship between the real - scene objects corresponding to each single - entity data based on the semantic information, and label the single - entity data based on the spatial position relationship to obtain the compositional features of the real - scene data;

[0028] Split the real - scene data based on the compositional features to obtain multiple groups of real - scene data groups, and extract the attribute information of each group of real - scene data groups based on the labeling results;

[0029] Obtain the scene classification name corresponding to each group of real - scene data groups based on the attribute information, and perform an affiliated label on the corresponding real - scene data group based on the scene classification name to obtain a set of modeling tasks.

[0030] Preferably, for a cloud - platform real - scene modeling control method, in step 3, distribute the set of modeling tasks to each modeling node in a distributed manner, and perform dynamic interaction analysis on each modeling task based on the modeling node, and couple and coordinate the analysis results to construct a real - scene model, including:

[0031] Obtain the system composition of the real - scene modeling, and perform node division on the system composition based on the functional attributes to obtain a set of modeling nodes;

[0032] Determine the mapping relationship between the modeling tasks and the modeling nodes based on the functional attributes of the modeling nodes and the task requirements of each modeling task in the set of modeling tasks, construct a distributed distribution link based on the mapping relationship, and parallelly distribute each modeling task to each modeling node based on the distributed distribution link;

[0033] Control each modeling node to perform task parsing on the corresponding modeling task based on the parallel distribution result to obtain the frame structure corresponding to each modeling task, and perform task verification on the modeling task based on the frame structure;

[0034] Based on the interactive mechanism of modeling nodes, the modeling task is partially modified according to the task verification result, and the structure of the partially modified modeling task is analyzed based on the modeling nodes to obtain the local structural features and three-dimensional point cloud data of the real scene;

[0035] Obtaining edge contours of the real scene based on local structural features, and constructing a model framework of the real scene based on the edge contours; at the same time, determining a hierarchical structure of the real scene based on the edge contours, and sequentially constructing a local three-dimensional spatial structure in the model framework according to the three-dimensional point cloud data based on the hierarchical structure;

[0036] Directed association is performed on adjacent local three-dimensional spatial structures, and based on the directed association result, the local three-dimensional spatial structure is nested and associated with the model framework of the real scene;

[0037] Performing texture mapping on the nested association results based on the real scene data to obtain an initial real scene model of the real scene;

[0038] Determine the spatial position of each initial real scene model based on the real scene data, and determine the structural splicing boundary of adjacent initial real scene models based on the spatial position;

[0039] The initial real-scene models of each modeling node are structurally spliced ​​based on the structural splicing boundary, and the structural splicing results are coupled and coordinated to obtain the final real-scene model.

[0040] Preferably, a cloud platform reality modeling control method is provided, which sends each modeling task to each modeling node in parallel based on a distributed sending link, including:

[0041] Based on the parallel delivery results, the start time of each modeling node is recorded. At the same time, the background record log is built and the record log is divided into regions;

[0042] Based on the regional division results, the start time of each modeling node is entered as initial record data, and based on the entry results, the real-time modeling data of each modeling node is synchronously recorded;

[0043] Perform periodic traversal on the input results in the record log based on a preset time interval, and obtain the progress index of each modeling node under the start time limit based on the periodic traversal results;

[0044] Determine the modeling status of each modeling node based on the relative size relationship between the progress indicator and the preset progress indicator, and lock the abnormal modeling node based on the modeling status;

[0045] Adjust the parameters of the abnormal modeling nodes until the preset progress indicator requirements are met.

[0046] Preferably, for a cloud platform real - scene modeling control method, in step 4, multi - terminal visualization configuration is performed on the real - scene model to complete real - scene modeling, including:

[0047] Obtain the obtained real - scene model, determine the available terminals of the real - scene model, and extract the operating environments of the available terminals;

[0048] Based on the operating environment, determine the development frameworks of each available terminal, and perform model parameter conversion on the real - scene model based on the development frameworks;

[0049] Based on the model parameter conversion results, determine the interface layout of the real - scene model on each available terminal, and configure the resolution of the interface layout of each available terminal based on the basic configuration of each available terminal;

[0050] Based on the configuration results, test the real - scene model on each available terminal, and perform optimized rendering on the real - scene model on each available terminal based on the test results to complete real - scene modeling.

[0051] The present invention provides a cloud platform real - scene modeling control system, including:

[0052] A data acquisition module, configured to collect real - scene data based on a preset shooting mode by a drone and upload the real - scene data to the cloud platform;

[0053] A task determination module, configured to extract the composition characteristics of the real - scene data based on the cloud platform, and split the real - scene data based on the composition characteristics to obtain a set of modeling tasks;

[0054] A real - scene model construction module, configured to distribute the set of modeling tasks to each modeling node in a distributed manner, perform dynamic interaction analysis on each modeling task based on the modeling node, and perform coupling coordination on the analysis results to construct a real - scene model;

[0055] A visualization configuration module, configured to perform multi - terminal visualization configuration on the real - scene model to complete real - scene modeling.

[0056] Preferably, for a cloud platform real - scene modeling control system, the data acquisition module includes:

[0057] A parameter configuration unit, configured to obtain the real - scene data acquisition requirements, and analyze the real - scene data acquisition requirements to obtain the acquisition transformation angle and acquisition accuracy of the oblique camera on the drone;

[0058] A route determination unit, configured to simultaneously obtain the scene distribution characteristics of the area to be collected, and determine the flight route of the drone over the area to be collected based on the scene distribution characteristics and the acquisition accuracy;

[0059] The data acquisition unit is used to obtain the preset shooting mode of the drone based on the acquired change angle and flight route, and control the drone to perform real-scene data acquisition based on the preset shooting mode.

[0060] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0061] 1. By controlling the drone to collect real-scene data according to the preset shooting mode, the comprehensive and effective acquisition of real-scene data is realized, and the collected real-scene data is uploaded to the cloud platform, which is convenient for the cloud platform to process the real-scene data and provides reliable data support for real-scene modeling. Secondly, by splitting the real-scene data through the cloud platform, the accurate and effective determination of the modeling task set is realized, and the modeling tasks are distributed to each modeling node in a distributed manner, which is convenient for each modeling node to synchronously execute the modeling tasks, improves the real-scene modeling efficiency, and during the real-scene modeling, the dynamic interaction and analysis between the modeling nodes ensure the accuracy and reliability of the models finally obtained by each modeling node. Finally, the results obtained by each modeling node are coupled and coordinated to realize the accurate and effective construction of the real-scene model, and the multi-terminal visualization configuration of the real-scene model is carried out to ensure that the real-scene model can operate effectively on multiple different terminals, improving the effect and accuracy of the real-scene modeling.

[0062] 2. By analyzing the requirements for real-scene data acquisition, the accurate and effective determination of the acquisition change angle and acquisition accuracy of the tilt camera on the drone is realized. Secondly, by determining the scene distribution characteristics of the area to be collected, the flight route of the drone over the area to be collected is locked according to the scene distribution characteristics, which is convenient for the drone to perform comprehensive and effective real-scene data acquisition. Finally, the acquisition change angle and flight route are summarized to effectively determine the preset shooting mode of the drone, providing a reliable guarantee for controlling the drone to perform real-scene data acquisition and ensuring the accuracy and reliability of the collected real-scene data.

[0063] Other features and advantages of the present invention will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present invention. The objectives and other advantages of the present invention can be achieved and obtained through the structures specifically pointed out in this application document.

[0064] The technical solutions of the present invention will be further described in detail below through the drawings and embodiments. Description of the Drawings

[0065] The drawings are used to provide a further understanding of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the present invention, and do not constitute a limitation to the present invention. In the drawings:

[0066] Figure 1Flow chart of a cloud platform real - scene modeling control method in an embodiment of the present invention;

[0067] Figure 2 Flow chart of step 1 in a cloud platform real - scene modeling control method in an embodiment of the present invention;

[0068] Figure 3 Structure diagram of a cloud platform real - scene modeling control system in an embodiment of the present invention. Detailed implementation manners

[0069] The following is a description of the preferred embodiments of the present invention with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are only for the purpose of illustrating and explaining the present invention, and are not used to limit the present invention.

[0070] Embodiment 1:

[0071] This embodiment provides a cloud platform real - scene modeling control method. As Figure 1 shown, it includes:

[0072] Step 1: Collect real - scene data based on a preset shooting mode by a drone and upload the real - scene data to the cloud platform;

[0073] Step 2: Extract the composition features of the real - scene data based on the cloud platform, and split the real - scene data based on the composition features to obtain a set of modeling tasks;

[0074] Step 3: Distribute the set of modeling tasks to each modeling node in a distributed manner, perform dynamic interaction analysis on each modeling task based on the modeling node, and couple and coordinate the analysis results to construct a real - scene model;

[0075] Step 4: Perform multi - terminal visualization configuration on the real - scene model to complete real - scene modeling.

[0076] In this embodiment, the preset shooting mode is set in advance, including zigzag flight, circular flight, and the shooting angle of the oblique camera carried on the drone.

[0077] In this embodiment, the real - scene data refers to the parameters or images that can represent the on - site scene situation obtained by collecting the specific situation on - site by a drone.

[0078] In this embodiment, the composition features refer to the data types included in the real - scene data. For example, they can be real - scene data about roads, buildings, forests, etc.

[0079] In this embodiment, the set of modeling tasks refers to different types of modeling data obtained by splitting the real - scene data according to the composition features of the real - scene data, including modeling tasks corresponding to buildings, modeling tasks for roads, and modeling tasks for rivers, etc.

[0080] In this embodiment, the modeling nodes are pre-set and are the execution entities for constructing different parts of the scene model, and are a part of the entire modeling system.

[0081] In this embodiment, dynamic interaction analysis refers to the modeling nodes verifying each modeling task and, when there are differences in the modeling tasks, promptly correcting the modeling tasks.

[0082] In this embodiment, coupling and coordination refer to associating and splicing the local models finally obtained by different modeling nodes, and finally realizing the acquisition of the real-scene model.

[0083] In this embodiment, multi-terminal visualization configuration refers to configuring the format, resolution, and interaction method of the real-scene model, with the aim of ensuring that the real-scene model can run effectively on different terminals.

[0084] The working principle and beneficial effects of the above technical solution are as follows: By controlling the drone to collect real-scene data according to the preset shooting mode, a comprehensive and effective acquisition of the real-scene data is achieved, and the collected real-scene data is uploaded to the cloud platform, facilitating the cloud platform to process the real-scene data, providing reliable data support for real-scene modeling. Secondly, by splitting the real-scene data through the cloud platform, an accurate and effective determination of the modeling task set is realized, and the modeling tasks are distributed to each modeling node, facilitating each modeling node to synchronously execute the modeling tasks, improving the real-scene modeling efficiency. When performing real-scene modeling, dynamic interaction analysis is carried out among the modeling nodes to ensure the accuracy and reliability of the models finally obtained by each modeling node. Finally, the results obtained by each modeling node are coupled and coordinated to accurately and effectively construct the real-scene model, and multi-terminal visualization configuration is performed on the real-scene model to ensure that the real-scene model can run effectively on multiple different terminals, improving the effect and accuracy of real-scene modeling.

[0085] Embodiment 2:

[0086] Based on Embodiment 1, this embodiment provides a cloud platform real-scene modeling control method. As Figure 2 shown, in step 1, collecting real-scene data based on the drone according to the preset shooting mode includes:

[0087] Step 101: Obtain the real-scene data collection requirements, and analyze the real-scene data collection requirements to obtain the collection transformation angle and collection accuracy of the oblique camera on the drone.

[0088] Step 102: At the same time, obtain the scene distribution characteristics of the area to be collected, and determine the flight route of the drone above the area to be collected based on the scene distribution characteristics and the collection accuracy.

[0089] Step 103: Based on the collected change angle and flight route, determine the preset shooting mode of the UAV, and control the UAV to collect real-scene data based on the preset shooting mode.

[0090] In this embodiment, the requirements for real-scene data collection are set in advance, including the collection angle, collection accuracy, etc.

[0091] In this embodiment, the collection change angle refers to the angle change of each collection by the oblique camera on the UAV, aiming to collect real-scene data at different angles.

[0092] In this embodiment, the scene distribution feature refers to the area distribution of the area to be collected, so as to facilitate the determination of the flight route of the UAV.

[0093] The working principle and beneficial effects of the above technical solution are as follows: By analyzing the requirements for real-scene data collection, accurately and effectively determine the collection change angle and collection accuracy of the oblique camera on the UAV. Secondly, determine the scene distribution feature of the area to be collected, and lock the flight route of the UAV over the area to be collected according to the scene distribution feature, which is convenient for the UAV to conduct comprehensive and effective real-scene data collection. Finally, summarize the collection change angle and flight route to effectively determine the preset shooting mode of the UAV, providing a reliable guarantee for controlling the UAV to collect real-scene data and ensuring the accuracy and reliability of the collected real-scene data.

[0094] Embodiment 3:

[0095] Based on the embodiment 2, this embodiment provides a cloud platform real-scene modeling control method, which controls the UAV to collect real-scene data based on the preset shooting mode, including:

[0096] Obtain the collected real-scene data, perform edge contour detection on the real-scene images of the real-scene data, and obtain the edge contour bounding rectangle of the real-scene image at each collection angle;

[0097] Determine adjacent image groups of the real-scene images of the real-scene data based on the flight route of the UAV over the area to be collected, and perform opposite overlap on the real-scene images in the adjacent image groups;

[0098] Determine the overlap degree of the edge contour bounding rectangles of adjacent real-scene images based on the opposite overlap, and when the overlap degree does not meet the preset threshold, re-collect the real-scene data;

[0099] Otherwise, retain the real-scene images in the adjacent image groups, and add position labels to the real-scene images based on the retention results.

[0100] In this embodiment, edge contour detection refers to detecting the edge distribution state of a real-scene image, including edge structures and shapes, etc.

[0101] In this embodiment, the edge contour enclosing rectangle refers to a border that can cover the edges of a real-scene image.

[0102] In this embodiment, adjacent image groups refer to the real-scene images corresponding to adjacent flight routes during the image acquisition by the drone, that is, the real-scene images with the same boundary area.

[0103] In this embodiment, opposite overlap refers to the overlap degree matching of adjacent real-scene images.

[0104] In this embodiment, the preset threshold is set in advance and can be adjusted.

[0105] In this embodiment, the position label refers to a marking symbol that can distinguish the positions of the scenes recorded in the real-scene images.

[0106] The working principle and beneficial effects of the above technical solution are as follows: By performing edge contour detection on the real-scene images of the collected real-scene data, the edge contour enclosing rectangle of the real-scene image is determined according to the edge contour detection result, which provides convenience for determining the overlap degree. Secondly, the adjacent image groups of the real-scene images are determined according to the flight route, and the overlap degree of the adjacent image groups is matched, so as to facilitate the effective judgment of the acquisition qualification of the real-scene data according to the overlap degree. Finally, when the overlap degree does not meet the preset threshold, the real-scene data is collected again. Otherwise, the collected real-scene data is retained, providing reliable data support for real-scene modeling.

[0107] Embodiment 4:

[0108] Based on Embodiment 1, this embodiment provides a cloud platform real-scene modeling control method. In step 1, uploading the real-scene data to the cloud platform includes:

[0109] Obtaining the communication parameters of the local device and docking the local device and the drone based on the communication parameters;

[0110] Caching the real-scene data collected by the drone on the local device based on the docking result, and building a communication link between the local device and the cloud platform based on the caching result;

[0111] Compressing the cached real-scene data and splitting the compressed real-scene data to obtain compressed data blocks;

[0112] Uploading the compressed data blocks to the cloud platform in sequence based on the communication link.

[0113] In this embodiment, the communication parameters refer to the communication parameters of the local device, including communication bandwidth, communication format requirements, etc.

[0114] In this embodiment, the local device refers to a device capable of communicating and docking with the drone.

[0115] In this embodiment, the compressed data block refers to the result obtained after splitting the data compression result of the real-scene data.

[0116] The working principle and beneficial effects of the above technical solution are as follows: By constructing a communication link between the local device and the drone and a communication link between the local device and the cloud platform, the real-scene data collected by the drone can be effectively uploaded to the cloud platform through the local device, which facilitates the analysis of the real-scene data by the cloud platform and provides convenience for real-scene modeling.

[0117] Embodiment 5:

[0118] Based on Embodiment 1, this embodiment provides a method for controlling real-scene modeling on a cloud platform. In step 2, based on the cloud platform, the composition features of the real-scene data are extracted, and the real-scene data is split based on the composition features to obtain a set of modeling tasks, including:

[0119] The obtained real-scene data is acquired, and the real-scene data is analyzed based on the scene modeling knowledge system to obtain the semantic information of each single-piece data in the real-scene data;

[0120] Based on the semantic information, the spatial position relationship between the real-scene objects corresponding to each single-piece data is determined, and the single-piece data is labeled based on the spatial position relationship to obtain the composition features of the real-scene data;

[0121] Based on the composition features, the real-scene data is split into multiple groups of real-scene data groups, and the attribute information of each group of real-scene data groups is extracted based on the labeling results;

[0122] Based on the attribute information, the scene classification name corresponding to each group of real-scene data groups is obtained, and the corresponding real-scene data groups are marked with attachments based on the scene classification name to obtain a set of modeling tasks.

[0123] In this embodiment, the scene modeling knowledge system is set in advance and includes the data types corresponding to the real-scene data and the object types represented.

[0124] In this embodiment, the single-piece data refers to the specific data information corresponding to each moment in the real-scene data.

[0125] In this embodiment, the semantic information refers to the specific meaning content corresponding to each single-piece data, including the position information of the real-scene object corresponding to the single-piece data, etc.

[0126] In this embodiment, annotating the monomer data according to the spatial position relationship means determining the corresponding monomer data according to the free position relationship of each real-scene object, so as to facilitate the splitting of the real-scene data.

[0127] In this embodiment, multiple groups of real-scene data groups refer to the results obtained after splitting the real-scene data.

[0128] In this embodiment, the attribute information refers to the real-scene object type corresponding to each group of real-scene data groups, the location information, etc.

[0129] In this embodiment, the scene classification names include "house", "road", "river", etc.

[0130] In this embodiment, the affiliated mark refers to marking the corresponding real-scene data group with the obtained scene classification name.

[0131] The working principle and beneficial effects of the above technical solution are as follows: By analyzing the real-scene data according to the scene modeling knowledge system, the composition characteristics of the real-scene data are accurately and effectively determined. Secondly, the real-scene data is split according to the composition characteristics, and the attribute information of each group of real-scene data groups obtained by splitting is determined. Finally, the scene classification name of each group of real-scene data groups is determined according to the attribute information, and the affiliated mark of the real-scene data group is realized according to the scene classification name, so as to accurately and effectively determine the modeling task set, which is convenient for distributing the real-scene data to the corresponding modeling nodes respectively and improving the efficiency of scene modeling.

[0132] Embodiment 6:

[0133] Based on Embodiment 1, this embodiment provides a cloud platform real-scene modeling control method. In step 3, the modeling task set is distributed to each modeling node in a distributed manner, and dynamic interaction analysis is performed on each modeling task based on the modeling node, and the analysis results are coupled and coordinated to construct a real-scene model, including:

[0134] Obtain the system composition of the real-scene modeling, and perform node division on the system composition based on the functional attributes to obtain a set of modeling nodes;

[0135] Determine the mapping relationship between the modeling tasks and the modeling nodes based on the functional attributes of the modeling nodes and the task requirements of each modeling task in the modeling task set, construct a distributed distribution link based on the mapping relationship, and distribute each modeling task to each modeling node in parallel based on the distributed distribution link;

[0136] Control each modeling node to perform task parsing on the corresponding modeling task based on the parallel distribution result, obtain the frame structure corresponding to each modeling task, and perform task verification on the modeling task based on the frame structure;

[0137] The interaction mechanism based on modeling nodes makes local corrections to the modeling tasks according to the task verification results, and performs structural analysis on the locally corrected modeling tasks based on the modeling nodes to obtain the local structural features of the real scene and the three-dimensional point cloud data;

[0138] Based on the local structural features, the edge contour of the real scene is obtained, and the model framework of the real scene is constructed based on the edge contour. At the same time, the hierarchical structure of the real scene is determined based on the edge contour, and the local three-dimensional space structure is sequentially constructed in the model framework according to the three-dimensional point cloud data based on the hierarchical structure;

[0139] Directed association is performed on adjacent local three-dimensional space structures, and the local three-dimensional space structures are nested and associated with the model framework of the real scene based on the directed association results;

[0140] Texture mapping is performed on the nested association results based on the real scene data to obtain the initial real scene model of the real scene;

[0141] Based on the real scene data, the spatial positions of each initial real scene model are determined, and the structural splicing boundaries of adjacent initial real scene models are determined based on the spatial positions;

[0142] Based on the structural splicing boundaries, the initial real scene models of each modeling node are structurally spliced, and the structural splicing results are coupled and coordinated to obtain the final real scene model.

[0143] In this embodiment, the functional attribute refers to the functions or effects that can be achieved in different links of the real scene modeling system.

[0144] In this embodiment, the modeling node set refers to the links that can respectively execute the corresponding modeling requirements obtained by dividing the real scene modeling according to the functional attributes.

[0145] In this embodiment, the task requirement refers to the purpose that each modeling task needs to achieve and the types of modeling services that need to be carried out, etc.

[0146] In this embodiment, the mapping relationship refers to the corresponding relationship between the modeling task and the modeling node, so as to facilitate the distribution of the modeling task to the corresponding modeling node for corresponding real scene modeling operations.

[0147] In this embodiment, the framework structure refers to the basic framework of each real scene object obtained after each modeling node analyzes the modeling task.

[0148] In this embodiment, the task verification refers to verifying the framework structure determined by different modeling nodes, so as to facilitate determining that there are no duplicate or conflicting situations in the real scene modeling tasks performed by each modeling node.

[0149] In this embodiment, the interaction mechanism is pre-set and is a method for representing the communication or interaction between modeling nodes during the modeling operation.

[0150] In this embodiment, local correction refers to the adjustment of an abnormal modeling task when an abnormality occurs in the modeling task.

[0151] In this embodiment, local structural features refer to the specific structural conditions of the model parts that each modeling node needs to construct.

[0152] In this embodiment, the model framework is a framework structure that is constructed based on the edge contour of the actual scene and can represent the specific appearance form of the current part.

[0153] In this embodiment, the hierarchical structure refers to the hierarchical distribution corresponding to different heights in the actual scene, so as to facilitate the construction of a three-dimensional model consistent with the actual scene.

[0154] In this embodiment, the local three-dimensional space structure refers to the specific position distribution corresponding to the detail points inside and on the appearance in the actual scene determined according to the three-dimensional point cloud data.

[0155] In this embodiment, directed association refers to splicing the local three-dimensional space structures with position associations to ensure accurate filling of the model interior.

[0156] In this embodiment, nested association refers to placing the local three-dimensional space structures obtained by directed association in the model framework of the actual scene to achieve the effective construction of the scene model.

[0157] In this embodiment, texture mapping refers to filling the surface of the constructed model with textures and colors to ensure that the finally obtained model is consistent with the actual scene.

[0158] In this embodiment, the initial actual scene model refers to a part of the model in the actual scene finally obtained by each modeling node.

[0159] In this embodiment, the structural splicing boundary refers to the specific position for splicing adjacent initial actual scene models.

[0160] In this embodiment, coupling coordination refers to adjusting and optimizing the splicing result to ensure the integrity and wholeness of the finally obtained actual scene model.

[0161] The working principle and beneficial effects of the above technical solution are as follows: By dividing the nodes of the real-scene modeling system according to functional attributes, an accurate and effective determination of the modeling node set is achieved, and the mapping relationship between the modeling tasks and each modeling node is determined. Then, according to the mapping relationship, each modeling task is accurately and effectively distributed to the corresponding modeling node, improving the scene modeling efficiency. Secondly, through the parsing of the modeling tasks received by each modeling node, an accurate and effective construction of the initial real-scene models of each part in the real-scene is realized, ensuring the accuracy and reliability of the initial real-scene model of each part. Finally, according to the real-scene data, the spatial positions of each initial real-scene model are determined, and the structural splicing boundaries of the initial real-scene models corresponding to adjacent real-scene parts are determined according to the spatial positions, so as to facilitate the structural splicing of the initial real-scene models according to the structural splicing boundaries, ensuring the reliability of the finally obtained real-scene model and improving the effect and accuracy of real-scene modeling.

[0162] Embodiment 7:

[0163] Based on Embodiment 6, this embodiment provides a cloud platform real-scene modeling control method, which distributes each modeling task to each modeling node in parallel based on a distributed distribution link, including:

[0164] Based on the parallel distribution result, record the start time of each modeling node. At the same time, construct a background record log and divide the record log by region;

[0165] Based on the region division result, take the start time of each modeling node as the initial record data for entry, and synchronously record the real-time modeling data of each modeling node based on the entry result;

[0166] Periodically traverse the entry results in the record log at a preset time interval, and obtain the progress indicators of each modeling node under the start time limit based on the periodic traversal result;

[0167] Based on the relative magnitude relationship between the progress indicator and the preset progress indicator, determine the modeling status of each modeling node, and lock the abnormal modeling nodes based on the modeling status;

[0168] Adjust the parameters of the abnormal modeling nodes until the requirements of the preset progress indicator are met.

[0169] In this embodiment, the background record log is used to record the operation data when different modeling nodes execute modeling tasks.

[0170] In this embodiment, the region division refers to dividing the background record log according to the number of modeling nodes, so as to facilitate ensuring the one-to-one correspondence between the modeling nodes and the record regions.

[0171] In this embodiment, the initial recorded data refers to the time information when each modeling node starts to execute the modeling task, that is, the corresponding start time.

[0172] In this embodiment, the real-time modeling data refers to the specific execution data corresponding to each modeling node during the execution of the modeling process.

[0173] In this embodiment, the preset time interval is set in advance.

[0174] In this embodiment, the progress indicator refers to the progress of each modeling node in executing the modeling under different cycle traversals, including the modeling rate, etc.

[0175] In this embodiment, the preset progress indicator is set in advance, which is a reference basis for measuring whether the current progress indicator meets the requirements and can be adjusted.

[0176] In this embodiment, locking abnormal modeling nodes based on the modeling state means determining the modeling nodes whose progress indicators do not meet the requirements of the preset progress indicators as abnormal modeling nodes.

[0177] The working principle and beneficial effects of the above technical solution are as follows: By constructing a background record log and dividing the background record log into regions, it is convenient to effectively record the start time and real-time modeling data corresponding to different modeling nodes. Secondly, by traversing the record data in the background record log at preset time intervals and determining the modeling progress of different modeling nodes according to the traversal results, it is convenient to effectively determine the working status of different modeling nodes. Finally, based on the modeling status of each modeling node, abnormal modeling nodes are locked, and parameter adjustment is performed on the abnormal modeling nodes according to the locking results, ensuring the reliability and accuracy of modeling the real scene.

[0178] Embodiment 8:

[0179] Based on Embodiment 1, this embodiment provides a cloud platform real scene modeling control method. In step 4, multi-terminal visualization configuration is performed on the real scene model to complete real scene modeling, including:

[0180] Obtain the real scene model, determine the available terminals of the real scene model, and extract the operating environments of the available terminals;

[0181] Determine the development framework of each available terminal based on the operating environment, and perform model parameter conversion on the real scene model based on the development framework;

[0182] Determine the interface layout of the real scene model on each available terminal based on the model parameter conversion results, and configure the resolution of the interface layout of each available terminal based on the basic configuration of each available terminal;

[0183] Test the real - scene model on each available terminal based on the configuration results, and optimize and render the real - scene model on each available terminal based on the test results to complete real - scene modeling.

[0184] In this embodiment, the available terminal refers to a device on which the real - scene model can be successfully displayed, including computers, mobile phones, etc.

[0185] In this embodiment, the operating environment refers to the operating conditions and requirements of the available terminal.

[0186] In this embodiment, the development framework refers to the corresponding coding format or template when each available terminal is running or being built.

[0187] In this embodiment, the basic configuration refers to the performance parameters of each available terminal.

[0188] The working principle and beneficial effects of the above - mentioned technical solution are as follows: By determining the available terminals of the real - scene model and extracting the operating environment of the available terminals, parameter conversion of the real - scene model is realized according to the operating environment, ensuring that the real - scene model can be effectively visually displayed on different terminals. Secondly, the resolution of the interface layout of each available terminal is configured according to the basic configuration of each available terminal, ensuring the display effect of the real - scene model on each available terminal. Finally, the real - scene model is tested on different available terminals, ensuring the running effect of the real - scene model.

[0189] Embodiment 9:

[0190] This embodiment provides a real - scene modeling control system for a cloud platform, as Figure 3 shown, including:

[0191] A data acquisition module, which is used to collect real - scene data based on a drone according to a preset shooting mode and upload the real - scene data to the cloud platform;

[0192] A task determination module, which is used to extract the composition characteristics of the real - scene data based on the cloud platform and split the real - scene data based on the composition characteristics to obtain a set of modeling tasks;

[0193] A real - scene model construction module, which is used to distribute the set of modeling tasks to each modeling node in a distributed manner, perform dynamic interaction analysis on each modeling task based on the modeling node, and couple and coordinate the analysis results to construct a real - scene model;

[0194] A visualization configuration module, which is used to perform multi - terminal visualization configuration on the real - scene model to complete real - scene modeling.

[0195] The working principle and beneficial effects of the above technical solution are as follows: By controlling the drone to collect real-scene data according to the preset shooting mode, a comprehensive and effective acquisition of the real-scene data is achieved, and the collected real-scene data is uploaded to the cloud platform, which is convenient for the cloud platform to process the real-scene data and provides reliable data support for real-scene modeling. Secondly, by splitting the real-scene data through the cloud platform, an accurate and effective determination of the modeling task set is realized, and the modeling tasks are distributed to each modeling node in a distributed manner, which is convenient for each modeling node to synchronously execute the modeling tasks, improving the real-scene modeling efficiency. When performing real-scene modeling, dynamic interaction and analysis are carried out among the modeling nodes to ensure the accuracy and reliability of the models finally obtained by each modeling node. Finally, the results obtained by each modeling node are coupled and coordinated to accurately and effectively construct the real-scene model, and multi-terminal visualization configuration is performed on the real-scene model to ensure that the real-scene model can run effectively on multiple different terminals, improving the effect and accuracy of real-scene modeling.

[0196] Embodiment 10:

[0197] Based on Embodiment 9, this embodiment provides a cloud platform real-scene modeling control system. The data acquisition module includes:

[0198] A parameter configuration unit for obtaining the real-scene data acquisition requirements, parsing the real-scene data acquisition requirements, and obtaining the acquisition transformation angle and acquisition accuracy of the oblique camera on the drone;

[0199] A route determination unit for simultaneously obtaining the scene distribution characteristics of the area to be collected, and determining the flight route of the drone over the area to be collected based on the scene distribution characteristics and the acquisition accuracy;

[0200] A data acquisition unit for obtaining the preset shooting mode of the drone based on the acquisition change angle and the flight route, and controlling the drone to collect real-scene data based on the preset shooting mode.

[0201] The working principle and beneficial effects of the above technical solution are as follows: By parsing the real-scene data acquisition requirements, an accurate and effective determination of the acquisition transformation angle and acquisition accuracy of the oblique camera on the drone is realized. Secondly, by determining the scene distribution characteristics of the area to be collected, the flight route of the drone over the area to be collected is locked according to the scene distribution characteristics, which is convenient for the drone to comprehensively and effectively collect real-scene data. Finally, by summarizing the acquisition change angle and the flight route, an effective determination of the preset shooting mode of the drone is realized, providing a reliable guarantee for controlling the drone to collect real-scene data and ensuring the accuracy and reliability of the collected real-scene data.

[0202] Obviously, those skilled in the art can make various changes and modifications 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 cloud platform real scene modeling control method, characterized in that: include: Step 1: Collect real scene data based on the drone according to the preset shooting mode, and upload the real scene data to the cloud platform; Step 2: Extract the constituent features of the real scene data based on the cloud platform, and split the real scene data based on the constituent features to obtain a set of modeling tasks; Step 3: Distribute the modeling task set to each modeling node in a distributed manner, and dynamically interact and analyze each modeling task based on the modeling node, and couple and coordinate the analysis results to build a real scene model; Step 4: Perform multi-terminal visualization configuration on the real scene model to complete the real scene modeling; Among them, in step 3, the modeling task set is distributed to each modeling node, and each modeling task is dynamically interactively analyzed based on the modeling node, and the analysis results are coupled and coordinated to build a real scene model, including: Obtaining the system composition of the real scene modeling, and dividing the system composition into nodes based on functional attributes to obtain a modeling node set; Determine the mapping relationship between modeling tasks and modeling nodes based on the functional attributes of the modeling nodes and the task requirements of each modeling task in the modeling task set, build a distributed delivery link based on the mapping relationship, and deliver each modeling task to each modeling node in parallel based on the distributed delivery link; Based on the parallel delivery results, each modeling node is controlled to perform task analysis on the corresponding modeling task, the framework structure corresponding to each modeling task is obtained, and the modeling task is verified based on the framework structure; Based on the interactive mechanism of modeling nodes, the modeling task is partially modified according to the task verification result, and the structure of the partially modified modeling task is analyzed based on the modeling nodes to obtain the local structural features and three-dimensional point cloud data of the real scene; Obtaining edge contours of the real scene based on local structural features, and constructing a model framework of the real scene based on the edge contours; at the same time, determining a hierarchical structure of the real scene based on the edge contours, and sequentially constructing a local three-dimensional spatial structure in the model framework according to the three-dimensional point cloud data based on the hierarchical structure; Directed association is performed on adjacent local three-dimensional spatial structures, and based on the directed association result, the local three-dimensional spatial structure is nested and associated with the model framework of the real scene; Performing texture mapping on the nested association results based on the real scene data to obtain an initial real scene model of the real scene; Determine the spatial position of each initial real scene model based on the real scene data, and determine the structural splicing boundary of adjacent initial real scene models based on the spatial position; The initial real-scene models of each modeling node are structurally spliced ​​based on the structural splicing boundary, and the structural splicing results are coupled and coordinated to obtain the final real-scene model.

2. A cloud platform reality modeling control method according to claim 1, characterized in that: In step 1, real scene data is collected based on the drone according to the preset shooting mode, including: Obtain the real scene data collection requirements, and analyze the real scene data collection requirements to obtain the collection transformation angle and collection accuracy of the tilt camera on the drone; At the same time, the scene distribution characteristics of the area to be collected are obtained, and the flight route of the UAV over the area to be collected is determined based on the scene distribution characteristics and the collection accuracy; The preset shooting mode of the UAV is obtained based on the acquisition change angle and the flight route, and the UAV is controlled to collect real-scene data based on the preset shooting mode.

3. A cloud platform reality modeling control method according to claim 2, characterized in that: Control the drone to collect real-scene data based on preset shooting modes, including: Acquire the collected real scene data, and perform edge contour detection on the real scene image of the real scene data to obtain the edge contour enclosing rectangle of the real scene image at each collection angle; Determine adjacent image groups of real scene images of real scene data based on the flight path of the UAV over the area to be collected, and overlap the real scene images in the adjacent image groups in opposite directions; Determine the overlap degree of the edge contour enclosing rectangles of adjacent real scene images based on the opposite overlap, and re-collect the real scene data when the overlap degree does not meet a preset threshold; Otherwise, the real scene images in the adjacent image group are retained, and position tags are added to the real scene images based on the retained results.

4. A cloud platform reality modeling control method according to claim 1, characterized in that: In step 1, the real scene data is uploaded to the cloud platform, including: Obtain the communication parameters of the local device, and connect the local device and the drone based on the communication parameters; Based on the docking results, the real scene data collected by the drone is cached in the local device, and a communication link between the local device and the cloud platform is established based on the cached results; Compressing the cached real scene data, and splitting the compressed real scene data to obtain compressed data blocks; The compressed data blocks are uploaded to the cloud platform in sequence based on the communication link.

5. A cloud platform reality modeling control method according to claim 1, characterized in that: In step 2, the constituent features of the real scene data are extracted based on the cloud platform, and the real scene data is split based on the constituent features to obtain a set of modeling tasks, including: Acquire the obtained real scene data, and parse the real scene data based on the scene modeling knowledge system to obtain the semantic information of each monomer data in the real scene data; Determine the spatial position relationship between the real scene objects corresponding to each monomer data based on the semantic information, and annotate the monomer data based on the spatial position relationship to obtain the composition features of the real scene data; The real scene data is split based on the composition features to obtain multiple groups of real scene data, and attribute information of each group of real scene data is extracted based on the annotation results; The scene classification name corresponding to each group of real scene data is obtained based on the attribute information, and the corresponding real scene data group is labeled based on the scene classification name to obtain a modeling task set.

6. A cloud platform reality modeling control method according to claim 1, characterized in that: Based on the distributed delivery link, each modeling task is delivered to each modeling node in parallel, including: Based on the parallel delivery results, the start time of each modeling node is recorded. At the same time, the background record log is built and the record log is divided into regions; Based on the regional division results, the start time of each modeling node is entered as initial record data, and based on the entry results, the real-time modeling data of each modeling node is synchronously recorded; Perform periodic traversal on the input results in the record log based on a preset time interval, and obtain the progress index of each modeling node under the start time limit based on the periodic traversal results; Determine the modeling status of each modeling node based on the relative size relationship between the progress indicator and the preset progress indicator, and lock the abnormal modeling node based on the modeling status; Adjust the parameters of the abnormal modeling nodes until the preset progress indicator requirements are met.

7. A cloud platform reality modeling control method according to claim 1, characterized in that: In step 4, the real scene model is configured for multi-terminal visualization to complete the real scene modeling, including: Acquire the obtained real scene model, determine the available terminals of the real scene model, and extract the operating environment of the available terminals; Determine the development framework of each available terminal based on the operating environment, and convert model parameters of the real scene model based on the development framework; Determine the interface layout of the real scene model on each available terminal based on the model parameter conversion result, and configure the resolution of the interface layout of each available terminal based on the basic configuration of each available terminal; Based on the configuration results, the real scene model is tested on each available terminal, and based on the test results, the real scene model is optimized and rendered on each available terminal to complete the real scene modeling.

8. A cloud platform real scene modeling control system, characterized in that: include: A data acquisition module is used to collect real-scene data based on the drone according to a preset shooting mode and upload the real-scene data to the cloud platform; A task determination module is used to extract the composition features of the real scene data based on the cloud platform, and split the real scene data based on the composition features to obtain a modeling task set; The real scene model building module is used to distribute the modeling task set to each modeling node, and dynamically interact and analyze each modeling task based on the modeling node, and couple and coordinate the analysis results to build a real scene model; Visual configuration module, used to perform multi-terminal visual configuration of the real scene model to complete the real scene modeling; Among them, the real scene model construction module includes: Obtaining the system composition of the real scene modeling, and dividing the system composition into nodes based on functional attributes to obtain a modeling node set; Determine the mapping relationship between modeling tasks and modeling nodes based on the functional attributes of the modeling nodes and the task requirements of each modeling task in the modeling task set, build a distributed delivery link based on the mapping relationship, and deliver each modeling task to each modeling node in parallel based on the distributed delivery link; Based on the parallel delivery results, each modeling node is controlled to perform task analysis on the corresponding modeling task, the framework structure corresponding to each modeling task is obtained, and the modeling task is verified based on the framework structure; Based on the interactive mechanism of modeling nodes, the modeling task is partially modified according to the task verification result, and the structure of the partially modified modeling task is analyzed based on the modeling nodes to obtain the local structural features and three-dimensional point cloud data of the real scene; Obtaining edge contours of the real scene based on local structural features, and constructing a model framework of the real scene based on the edge contours; at the same time, determining a hierarchical structure of the real scene based on the edge contours, and sequentially constructing a local three-dimensional spatial structure in the model framework according to the three-dimensional point cloud data based on the hierarchical structure; Directed association is performed on adjacent local three-dimensional spatial structures, and based on the directed association result, the local three-dimensional spatial structure is nested and associated with the model framework of the real scene; Performing texture mapping on the nested association results based on the real scene data to obtain an initial real scene model of the real scene; Determine the spatial position of each initial real scene model based on the real scene data, and determine the structural splicing boundary of adjacent initial real scene models based on the spatial position; The initial real-scene models of each modeling node are structurally spliced ​​based on the structural splicing boundary, and the structural splicing results are coupled and coordinated to obtain the final real-scene model.

9. A cloud platform reality modeling control system according to claim 8, characterized in that: Data acquisition module, including: The parameter configuration unit is used to obtain the real scene data collection requirements, and analyze the real scene data collection requirements to obtain the collection transformation angle and collection accuracy of the tilt camera on the drone; A route determination unit is used to simultaneously obtain the scene distribution characteristics of the area to be collected, and determine the flight route of the UAV over the area to be collected based on the scene distribution characteristics and the collection accuracy; The data acquisition unit is used to obtain a preset shooting mode of the drone based on the acquired change angle and flight path, and control the drone to collect real-scene data based on the preset shooting mode.

Images