Method and system for generating high-precision three-dimensional live-action model

Abstract

The invention discloses a method and a system for generating a high-precision three-dimensional live-action model. The method comprises the following steps: acquiring data, constructing a foundation inclined live-action three-dimensional model and forming the three-dimensional live-action model. And editing the water area cavity based on the basic inclined live-action three-dimensional model to form a three-dimensional live-action model. The invention discloses a method for generating a high-precision three-dimensional live-action model, and the method comprises the steps: firstly carrying out the image collection of a target region, and then constructing a three-dimensional live-action model. In the model construction process, key processing work is carried out for the water area part, water area holes are effectively filled, and the integrity of the model is ensured. Therefore, the high-precision three-dimensional live-action model system can emphatically strengthen the dynamic monitoring and early warning capability of the surface change caused by rainfall, realizes accurate pre-judgment and timely early warning of the flooding range, and provides firm and reliable technical support and scientific decision basis for disaster prevention and reduction work.

Description

Technical Field

[0001] The present invention belongs to the field of geographic surveying and mapping technology, and relates to a method and system for generating a high-precision three-dimensional real scene model. Background Art

[0002] 3D reality models, as a realistic, three-dimensional, and time-series reflection of the spatiotemporal information of human production, living, and ecological spaces, can accurately depict the three-dimensional spatial location and attribute information of topography, land cover, and structures, providing refined spatial data support for urban planning, construction, and management. 3D reality models can more accurately assess land use, plan urban layout, and optimize resource allocation. By combining meteorological and geological data, the likelihood, scope, and intensity of disasters can be more accurately predicted, improving the accuracy and timeliness of early warnings.

[0003] In areas with widespread mountains and hills, geological disasters are frequent and concentrated due to the unique topography. Furthermore, the region's long rainy season and abundant rainfall make heavy rainfall highly susceptible to secondary disasters such as landslides, debris flows, and flash floods. However, the 3D real-life model contains a large amount of water, and the feature points of these areas are not obvious, making it extremely difficult for the software to extract these points. This, in turn, leads to holes in the reconstructed model, hindering the dynamic monitoring and early warning capabilities of rainfall-induced terrain changes, and making it impossible to accurately predict the extent of flooding and provide timely warnings. Summary of the Invention

[0004] In order to solve the above problems, the present invention adopts the following technical solutions:

[0005] A method for generating a high-precision three-dimensional real scene model, comprising the steps of:

[0006] Data collection: Deploy image control points on the surface of the area; take aerial photos of the entire surface of the area to obtain several images;

[0007] Constructing a basic tilted real-scene 3D model: Perform aerial triangulation based on the control point data, image data, and camera parameters to construct a basic tilted real-scene 3D model;

[0008] Forming a 3D real scene model: Editing the water cavity based on the basic tilted real scene 3D model to form a 3D real scene model; specifically including:

[0009] The Canny or Sobel algorithm is used to extract the boundaries of the waterside of the basic tilted real-life 3D model; semantic segmentation is performed on the boundaries to identify water areas, suspended areas below the ground, isolated water bodies with an area or volume below a threshold, and depressions as potential water areas;

[0010] In the basic tilted real-scene 3D model, select the water area and outline the boundary of the water area; select the central island on the water surface and outline the boundary of the central island; select the isolated water body whose area or volume is lower than the threshold and outline the boundary of the isolated water body; select the depression area and outline the boundary of the depression;

[0011] Based on the range lines formed by the boundary and the range line types, the water holes are reconstructed or filled until the image of the 3D reality model is complete.

[0012] As a further solution of the present invention: the reconstruction or filling of the water cavity includes:

[0013] Constrain the elevation of the water cavity to be consistent with the surrounding water surface, apply texture mapping technology to give the water a glossy effect; apply the fluid dynamics model to give the water cavity a dynamic effect;

[0014] Set a threshold for the area or volume of water holes to avoid filling real islands. After filling, verify whether the water holes are covered by plants.

[0015] As a further solution of the present invention: the step of arranging image control points on the surface of the area includes:

[0016] Deploy image control points: Deploy one image control point every 5 kilometers on the surface of the area;

[0017] Checkpoints: One checkpoint will be set up every 8 kilometers on the surface of the area;

[0018] Measuring image control points: Obtain the planar position of image control points based on PTK measurement or total station traverse measurement; obtain the elevation position of image control points based on geometric leveling, total station trigonometric height measurement, or RTK height measurement combined with a refined regional quasi-geoid model;

[0019] Combining the plane position and elevation position of each image control point, a combined map of the imaging control point distribution is drawn.

[0020] As a further solution of the present invention, the step of arranging image control points on the surface of the area further includes:

[0021] Each image control point shall be observed for at least 3 rounds, and the difference of the coordinate components of the plane position shall be less than 2cm, and the difference of the coordinate components of the elevation position shall be less than 3cm. If it exceeds the range, the number of measurements shall be increased; each detection point shall be observed for at least 1 round, and the difference of the coordinate components of the detection point shall be less than 5cm. If it exceeds the range, image control points shall be added at both ends of the detection point and measured; the median of each image control point shall be taken as the final observation result.

[0022] As a further solution of the present invention, the step of obtaining a plurality of images by aerial photography of the complete coverage of the surface in the area includes:

[0023] Set the flight difference mode: set to post-processing difference mode;

[0024] Flight routes are developed based on the terrain within the region, combined with environmental restrictions, flight constraints, and weather changes. The flight routes include the location relationship information of the departure point, en route points, and destination point, the flight altitude and speed, and the time period required to reach the destination. There are navigation edges between flight routes, with a heading overlap of 53% to 65%, with a minimum of no less than 53%. The relative heading overlap is 51% to 65%, and the heading overlap of adjacent image pairs is no less than 58%. The route edges exceed the range of the detection points and the range of the surveying and mapping area.

[0025] Flight data inspection: Perform quality inspection on flight data to ensure that the flight image data can effectively cover the area and ensure data quality; retake unqualified images; plan the retake route, and the two ends of the retake route should exceed a photographic baseline; for loopholes that do not affect the construction of the basic tilt real scene 3D model, retake only the loopholes;

[0026] Data processing: Analyze the resolution, photographic scale, ground resolution and image flight relationship of aerial photographs; organize and analyze camera data, image control point collection, and aerial photograph index maps; and perform post-differential flight count calculation on UAV data.

[0027] As a further solution of the present invention: the step of calculating the number of drone data by post-difference includes:

[0028] Calculate the location of base station data, mobile station data, base station coordinate information, antenna-camera relative position information, and base station instrument height information;

[0029] Correction of differential offset: Calculate the offset value of the camera in the NEZ direction relative to the antenna, with the antenna as the center. Calculate the position data of each point in the real-time differential data according to the azimuth at the time of flight, and correct the offset value to the position information corresponding to each camera shooting point.

[0030] As a further solution of the present invention: the step of constructing the basic tilted real scene three-dimensional model includes:

[0031] Use automated data checking tools to ensure the data quality of acquired images meets requirements;

[0032] Based on the original image and POS data, the project automatically connects the points and preliminarily determines the correspondence between different images. The project also uses the measured image control points to absolutely locate and orient the oblique image, correct the exterior orientation elements, restore the true position and attitude of each photo, complete the aerial triangulation, and generate the aerial triangulation report.

[0033] Automatic creation of tilted real-scene 3D models based on aerial triangulation reports;

[0034] Generate DSM and DOM results based on aerial triangulation reports.

[0035] As a further solution of the present invention: automatic creation of a tilted real-scene three-dimensional model based on an aerial triangulation report includes: retrieving the clearest images from each corresponding perspective for texture mapping; after the first texture mapping is completed, a color balancing algorithm is applied to the tiles with texture information, and an overall balance value is calculated; and the balance value is applied to the subsequent texture mapping process of the three-dimensional model.

[0036] As a further solution of the present invention, it also includes using lightweight tools to perform lightweight processing on the three-dimensional real scene model to optimize the model structure and texture.

[0037] A high-precision three-dimensional real scene model system is generated based on a method for generating a high-precision three-dimensional real scene model, comprising:

[0038] Functional modules with 2D data, 3D data and directory tree;

[0039] Oblique photography loading effect diagram;

[0040] Disaster warning zoning module: color rendering of different areas, setting transparency and elevation data;

[0041] Risk point and hidden danger point module: Different colors represent different types of disaster points, which are plotted on the 3D real-life model. Click the icon to display the measurement point information.

[0042] Terrain elevation profile analysis module: select points to draw profile lines, display profile elevation information in a graphical format, and obtain coordinate information by moving within the oblique photography loading effect diagram;

[0043] Terrain Contour Analysis Module: There are two areas to choose from for drawing, namely rectangle and polygon. The contour lines can be controlled and displayed by adjusting the interval and line width sliders on the panel.

[0044] Slope and aspect analysis module: There are three drawing methods: rectangle, polygon and point. Different drawing methods have different processing times. The larger the drawing area, the longer the waiting time.

[0045] Cut and fill analysis module: draw the cut and fill area and calculate the volume of the cut area;

[0046] Flood analysis module: Draw the flooded area, control the altitude and flooding speed; slide to select the flooding height to understand the flooded area.

[0047] Beneficial effects of the present invention:

[0048] A method for generating high-precision 3D real-world models begins by capturing images of the target area and then constructing a 3D real-world model. During the model construction process, key processing is performed on the water area to effectively fill in voids and ensure model integrity. This high-precision 3D real-world model system can enhance dynamic monitoring and early warning capabilities for rainfall-induced surface changes, enabling accurate prediction and timely warning of flooding extent, providing solid and reliable technical support and a scientific basis for decision-making in disaster prevention and mitigation efforts. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a flow chart of a method for generating a high-precision three-dimensional reality model;

[0050] Figure 2 It is a schematic diagram of a partial flow chart of step 1 of a method for generating a high-precision three-dimensional reality model;

[0051] Figure 3 A schematic diagram of another part of the process of step 1 of a method for generating a high-precision three-dimensional reality model;

[0052] Figure 4 It is a flowchart of step 2 of a method for generating a high-precision three-dimensional reality model;

[0053] Figure 5 It is a schematic diagram of step 3 of a method for generating a high-precision three-dimensional reality model;

[0054] Figure 6 This is a schematic diagram before the water hole is filled in step 3;

[0055] Figure 7 This is a schematic diagram after the water hole is filled in step 3;

[0056] Figure 8 It is a functional module diagram of a high-precision three-dimensional real-scene model system. DETAILED DESCRIPTION

[0057] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. It should be understood that this application is not limited to the example embodiments disclosed herein. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0058] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature identified as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, "plurality" means two or more, unless otherwise specifically defined.

[0059] In the embodiments of the present invention, unless otherwise expressly specified or limited, the terms "installed," "connected," "connected," "fixed," etc. should be understood in a broad sense. For example, they may refer to fixed connection, detachable connection, or integration; mechanical connection or electrical connection; direct connection or indirect connection through an intermediate medium; internal communication between two components or interaction between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0060] Example 1

[0061] like Figure 1-5 As shown, a method for generating a high-precision three-dimensional real scene model includes the following steps:

[0062] S1 Data Collection: Deploy image control points on the surface of the area; perform aerial photography to fully cover the surface of the area and obtain several images;

[0063] S2 builds a basic tilted real-scene 3D model: aerial triangulation is performed based on the data of control points, image data, and camera parameters to build a basic tilted real-scene 3D model;

[0064] S3 forms a three-dimensional real scene model: based on the basic inclined real scene three-dimensional model, the water cavity is edited to form a three-dimensional real scene model.

[0065] In this embodiment, if Figure 2 As shown, step S1 includes:

[0066] S111 deployment of image control points: one image control point is deployed every 5 kilometers on the surface within the area; deployment of checkpoints: one checkpoint is deployed every 8 kilometers on the surface within the area.

[0067] Specifically, to ensure data accuracy, image control points and checkpoints should be appropriately set. Image control points should meet the following requirements: Image control points should generally be located within the overlap of six or five images in both heading and lateral directions. The target image for image control points should be clear and easily distinguishable. When the target image conflicts with other image conditions, the target condition should be prioritized. Image control points should be located near the centerline of lateral overlap, at least 3 cm (23 cm x 23 cm image width) from the azimuth line. Separate points should be placed when lateral overlap is excessive. If lateral overlap is too small, the vertical distance between two points should generally be less than 1 cm on the photo, and no more than 2 cm when difficult. Image control points should generally be no less than 1.5 cm from the edge of the photo and no less than 1 mm from various photo markers. Image control points at the junction of different operating areas should be located at the overlap of the flight lines, with adjacent areas sharing as much as possible. If sharing is not possible, separate points should be placed.

[0068] Specifically, the layout of image control points fully utilizes IMU / GNSS-assisted aerial photography technology. In accordance with current specifications and accuracy requirements, the regional grid layout method is used. Image control points are laid out and collected in combination with the actual needs of the project and aerial photography planning. It is expected that one image control point will be laid out every 5 kilometers. Slopes with sharp changes in elevation should not be used as target points for selection. When image control points are inserted at the edge of vegetation, high ground or steep slopes, it should be stated whether the point falls above or below the step, and the relative height of the point to the larger reference ground should be measured to 0.1m. It should also be stated where the point is inserted and where the elevation is measured, and noted on the back of the photo. Stakes or marks should be driven into the field, and photos of the points should be taken when necessary. The selected target points should be convenient for observation by GPS or total station.

[0069] Specifically, to verify the accuracy of internal encryption points and topographic maps, appropriate checkpoints should be established. These checkpoints should be located at prominent features and away from regional network image control points. The results of confidentiality points should be separately collated and submitted to the quality inspection department. Checkpoints should be evenly spaced 8 kilometers apart within the regional network as needed.

[0070] S112: Measure image control points: Obtain the planar position of image control points based on PTK measurement or total station traverse measurement; obtain the elevation position of image control points based on geometric leveling, total station trigonometric height measurement, or RTK height measurement combined with a refined regional quasi-geoid model;

[0071] Specifically, high-precision measurement of image control points (GCPs) is performed to ensure accurate geographic correction of image coordinates during data processing. Operations are performed using single-frequency or dual-frequency GNSS receivers. Instrument parameters and antenna height information are checked and correctly set before each operation.

[0072] When measuring the plane of image control points, RTK full-field surveying is preferred. In areas where GPS is unsuitable, traverse and point-based surveying can be used. When measuring the elevation of image control points, RTK elevation measurement with quasi-geoid model refinement is used.

[0073] Each image control point is measured in at least three rounds, with the coordinate component error of the plane position less than 2cm and the coordinate component error of the elevation position less than 3cm. If the error exceeds the range, the number of measurements is increased. Each detection point is measured in at least one round, with the coordinate component error of the detection point less than 5cm. If the error exceeds the range, image control points are added at both ends of the detection point and measured. The median of each image control point is taken as the final observation result. The image control point data is calibrated multiple times to ensure its accurate correspondence with the actual geographic location. The mean error of the plane position of the plane control point and the horizontal elevation control point relative to the nearest basic control point is no more than 5cm. The mean error of the elevation control point and the horizontal elevation control point relative to the nearest basic control point should not be greater than 5cm.

[0074] S113: Drawing a combined map of imaging control point distribution based on the plane position and elevation position of each imaging control point.

[0075] Specifically, data processing for the inspection points involves using the GPS instrument's accompanying software to promptly import the raw data collected in the field from the data collector into a computer. The software then exports a measurement results table containing information such as point number, plane coordinates, elevation, horizontal accuracy, vertical accuracy, and number of satellites. Based on the measurement results table, a combined imaging control point layout diagram is drawn.

[0076] In this embodiment, if Figure 3 As shown, step S1 also includes:

[0077] S121: Set the flight difference mode: Set to post-processing difference mode.

[0078] Specifically, the machine's high-precision POS information is obtained using the PPK post-processing difference method. The operating steps of the PK post-processing difference operation mode are as follows:

[0079] Set up the GNSS receiver and align and level it;

[0080] Collect base station coordinates and measure instrument height;

[0081] Change the base station operation mode to static mode;

[0082] Download the static data file after the aerial survey mission is completed.

[0083] The static data collection intervals for mainstream receivers are 0.2 seconds, 0.5 seconds, 1 second, 2 seconds, 5 seconds, and 10 seconds. Aerial surveys typically use a 0.2-second interval. Set the host antenna height: Vertical, Slanted, or Pole. Select the measurement type and enter the measured height.

[0084] S122: Develop a flight route based on the terrain within the area, combined with environmental restrictions, flight constraints, and weather changes. The flight route includes the location relationship information of the departure point, transit points, and destination point, the flight altitude and speed, and the time period required to reach the destination. There are navigation edges between flight routes, with a heading overlap of 53% to 65%, with a minimum of no less than 53%. The relative heading overlap is 51% to 65%, and the heading overlap of adjacent image pairs is no less than 58%. The route edge exceeds the range of the detection point and the range of the surveying and mapping area.

[0085] Specifically, the entire area is covered, allowing drones to capture high-definition images of the surface using fixed-wing aircraft equipped with five-lens tilting cameras within their flight range. These images, processed by software, will be used for subsequent 3D modeling and analysis. The area is divided into multiple work zones. To ensure effective modeling, each zone must have a consistent flight path between them. During data collection, multiple flights, segmented flights, and recording at different time periods were used to ensure image clarity and coverage.

[0086] Route planning consists of two steps: first, pre-flight planning, which involves developing an optimal reference path based on the mission, environmental constraints, and flight restrictions. Second, in-flight re-planning involves dynamically adjusting the flight path or maneuvering tasks based on unexpected circumstances encountered during flight, such as terrain, weather changes, and unknown flight restrictions. Route planning includes the locational relationships between the departure point, en route points, and destination, as well as the flight altitude and speed and the timeframe required to reach the destination.

[0087] Environmental Restrictions: Missions may be subject to complex geographical constraints, such as military-controlled areas, obstacles, and difficult terrain. Therefore, flight operations should avoid these areas as much as possible. These areas can be marked as no-fly zones on maps to improve efficiency. Furthermore, weather factors within the flight area will also affect mission efficiency, so consider weather forecasts and response mechanisms for complex weather conditions such as strong winds, rain, and snow.

[0088] Flight constraints: The flight is performed using a drone, so the physical limitations of the drone impose the following constraints on the flight trajectory:

[0089] Minimum turning radius: Since the arc formed by the drone's flight turning will be limited by its own flight performance, it limits the drone to turning within a specific turning radius.

[0090] Maximum pitch angle: limits the track to turn within the vertical radius.

[0091] Minimum Track Segment Length: A drone's flight path consists of several segments between waypoints. During each segment, the drone flies in a straight line, but may change its flight attitude at certain waypoints based on mission requirements. The minimum track segment length specifies the minimum distance a drone must fly directly before it can begin changing its flight attitude.

[0092] Minimum safe flight altitude: limits the minimum flight altitude through the mission area to prevent the flight altitude from being too low and hitting the ground, resulting in a crash.

[0093] The specific flight missions that the drone must perform include arrival time and target approach direction. These requirements must be met: The track distance must be completed, limiting the track length to no more than a pre-set maximum distance. A fixed target approach direction ensures the drone approaches the target from a specific angle.

[0094] Replanning: On the one hand, when complete and accurate environmental information is available, the optimal trajectory from the starting point to the end point can be planned in one go. However, in reality, it is difficult to guarantee that the environmental information obtained will not change. On the other hand, due to mission uncertainty, drones often need to temporarily change their flight missions. When the environmental change area is small, the trajectory can be replanned online through local updates. However, when the environmental change area is large, drone mission planning must have online replanning capabilities.

[0095] S123: Flight data check: Perform a quality check on the flight data to ensure that the flight image data can effectively cover the area and ensure data quality; retake unqualified images; plan the retake route, and the two ends of the retake route should exceed a photographic baseline; for loopholes that do not affect the construction of the basic tilt real scene 3D model, retake only the loopholes;

[0096] Specifically, during the flight mission, the flight control management system is used for navigation control, and the system control software is used to control and monitor the real-time data collection. The following points should be noted during the data collection process.

[0097] Altitude maintenance: The altitude difference between adjacent images on the same route should not exceed 30m, the difference between the maximum and minimum altitudes should not exceed 50m, and the difference between the actual altitude and the designed altitude should not exceed 50m. When the relative altitude is less than or equal to 1000m, the difference between the actual altitude and the designed altitude within the aerial photography area should not exceed 50m. When the relative altitude is greater than 1000m, the difference between the actual altitude and the designed altitude should not exceed 5% of the designed altitude.

[0098] Flight speed: The flight speed should be kept as consistent as possible throughout the entire operating area; within a route, the aircraft's ascent and descent speeds should not vary by more than 10m / s.

[0099] Flight process: The pitch angle and roll angle of the route are generally no more than 2°, and the maximum does not exceed 4°; when the aircraft turns, the bank angle is no more than 15°; the route curvature is no more than 3%.

[0100] Make-up flight or re-flight: Use the aerial camera of the previous aerial photography flight to make up the flight.

[0101] Vulnerability re-photography should be performed according to the original design requirements. The spatial line connecting two adjacent photography stations along the route is called the photography baseline. Both ends of the re-photography route should extend beyond a photography baseline. For relative vulnerabilities that do not affect model construction, re-photograph only at the vulnerabilities.

[0102] Use dedicated software to perform quality checks on daily flight data to ensure that the imagery effectively covers the survey area and maintains data quality. We also conduct checks on other relevant raw data, such as POS data, to ensure the validity and integrity of the flight data. Comprehensive assessments and checks are conducted based on data integrity, accuracy requirements, and internal errors.

[0103] S124: Aerial data processing: Analyze the resolution, photographic scale, ground resolution, and image strip relationship of aerial photographs; organize and analyze camera data, image control point collection, and aerial photograph index maps; and perform post-differential flight count calculation on UAV data.

[0104] Specifically, data collation is a crucial step in the early stages of photogrammetry production. Correctly understanding the raw data significantly impacts the accuracy and effectiveness of the results. This phase involves analyzing the resolution, scale, ground resolution, and imagery's flight path relationships. It also involves analyzing and collcating camera files, control point files, and photo index maps.

[0105] Among them, the steps for solving the post-differential flight count of drone data include: solving the location of base station data, the location of mobile station data, base station coordinate information, antenna-camera relative position information, and base station instrument height information.

[0106] The drone data is solved by using the software's flight calculation function. It supports batch post-differential calculation of multiple flights and automatic identification of base station coordinates. Base station instrument height and antenna-camera phase difference information can be directly corrected in the differential calculation. Flight calculation instructions are as follows:

[0107] Base station data: Enter the base station observation file path.

[0108] Add mobile station data: Enter the mobile station observation file path.

[0109] Base station coordinates: Set the GNSS base station coordinate system to the geodetic coordinate system in the format of latitude and longitude + ellipsoidal height. Set the coordinate system to the projected coordinate system in the format of north-east height. Fill in the base station coordinate information. If the latitude and longitude are used, specify the latitude and longitude unit format. If the coordinates are plane coordinates, the default unit is meters.

[0110] Antenna-Camera Relative Position Information: The relative position of the drone's camera and the PPK board. The positive ENZ direction corresponds to the right-hand coordinate system. That is, N is the direction of the aircraft's nose, and Z is positive when it points upward. If the aircraft is a Southern system drone, quickly fill in the NEZ value by specifying the aircraft type and lens type.

[0111] Base station instrument height: Base station instrument height supports four filling methods: directly fill in the straight height or calculate the straight height by measuring the oblique height or measuring the height and side piece radius.

[0112] Solution Status: Set solution parameters, including the coordinate system information of the static station, the coordinate system information of the output differential file, the output accuracy, and the output latitude and longitude unit format. Display information such as the progress and problems during the solution process.

[0113] After the solution is completed, a solution report is generated.

[0114] Aerial surveys employ real-time differential analysis, requiring offset correction of the differential results. Real-time differential data records the position of the antenna phase center, while the final data required represents the position of each camera point. As shown in the figure below, the antenna and camera are offset, requiring correction of the real-time differential data for the antenna-camera offset.

[0115] S125: Differential offset correction step: Calculate the offset value of the camera in the NEZ direction relative to the antenna, with the antenna as the center. Calculate the position data of each point in the real-time differential data according to the azimuth during flight, and correct the offset value to the position information corresponding to each camera shooting point.

[0116] S126: Photo Arrangement: This software organizes the acquired data using the device connection function. It supports automatic grouping of multiple-flight photos, automatic identification of ground points and waste films, and one-click clearing of ground points and waste films from all lenses. It also supports handling situations such as lost films and lost points, provides interpolation and marking jump film tools, and comprehensively handles all data anomalies.

[0117] Step S1 can effectively reduce the impact of image distortion or position error, improve overall accuracy and reliability, and lay a solid foundation for subsequent 3D modeling and data analysis.

[0118] In this embodiment, if Figure 4As shown, step S2 includes:

[0119] S210: Use automated data checking tools to ensure that the data quality of the collected images meets the requirements;

[0120] S220: Automatically connect points based on the original image and POS data, preliminarily determine the correspondence between different images; use the measured image control points to absolutely locate and orient the oblique image, correct the exterior orientation elements, restore the true position and posture of each photo, complete the aerial triangulation, and generate an aerial triangulation report;

[0121] S230: Automatically create a tilted real scene three-dimensional model based on the aerial triangulation report;

[0122] When the modeling area is too large to output the model all at once, a segmented output method is adopted to determine a unified coordinate origin and the same distribution framework to stitch the segments together. Specifically, the geometric center of the region is selected as the global coordinate origin to establish a unified spatial reference system (including projection coordinate system and elevation benchmark) covering the entire modeling range; based on the hardware processing capabilities and model accuracy requirements, the modeling area is divided into several regular grid units to ensure that a 20% overlap is retained between each block; under the unified coordinate framework, a distributed computing framework is used to generate three-dimensional models of each block in parallel; through feature point matching algorithms, such as the SIFT+ICP hybrid algorithm, geometric alignment of the block models is achieved, and mask fusion technology is used to eliminate texture differences at the seams, ultimately generating a seamless complete model.

[0123] The clearest image from each corresponding perspective is retrieved for texture mapping. After the initial texture mapping is completed, a color balancing algorithm is applied to the tiles with texture information to calculate an overall balance value. This balance value is then applied to the subsequent texture mapping of the 3D model. Specifically, the aerial triangulation report is loaded into the modeling software, which automatically produces the oblique 3D model. Based on the stereo model and the corresponding aerial triangulation data, the software automatically retrieves the clearest image from each corresponding perspective for each face of the model for initial texture mapping. During flight, due to varying lighting, the oblique images from different perspectives may have color imbalances. This can be manually adjusted before the model's color is modified using the software. Specifically, the clearest image undergoes a preliminary manual color adjustment for texture mapping. After the initial texture mapping is completed, a color balancing algorithm is applied to the tiles with texture information, incorporating all processed textured tiles into the calculation. An overall balance value is calculated and used for subsequent production. Color balancing resets the reference 3D model texture, and the texture changes are applied to all newly submitted production results.

[0124] S240: Generate DSM and DOM results based on the aerial triangulation report.

[0125] The step S2 ensures that the final 3D model has high spatial consistency and color consistency.

[0126] In this embodiment, if Figure 5 As shown, step S3 includes:

[0127] S310: Use the Canny or Sobel algorithm to extract the boundaries of the waterside of the basic inclined real-life 3D model; perform semantic segmentation on the boundaries to identify water areas, suspended areas below the ground, isolated water bodies with an area or volume below a threshold, and depressions as potential water areas.

[0128] Specifically, an algorithm is used to automatically identify the waterside boundary and classify and label the boundary.

[0129] S320: In the basic inclined real-scene 3D model, a water area is selected and the boundary of the water area is outlined; a central island on the water surface is selected and the boundary of the central island is outlined; an isolated water body whose area or volume is lower than a threshold is selected and the boundary of the isolated water body is outlined; a depression is selected and the boundary of the depression is outlined.

[0130] Specifically, manually outlined images are used to further accurately determine the scope of the water area.

[0131] For closed holes, automatic identification algorithms are more effective, identifying depressions as potential water areas and subsequently filling them. To prevent misidentification, areas are used as thresholds, whether the location is within water, or elevation consistency (i.e., whether the hole is at water level) to distinguish between true islands and false holes. For connected holes or holes with long boundaries, manual refinement is used.

[0132] S330: Reconstruct or fill water holes according to the range lines formed by the boundary and the range line types until the image of the three-dimensional real scene model is complete.

[0133] Among them, reconstruction or filling of water holes includes:

[0134] Constrain the elevation of the water cavity to be consistent with the surrounding water surface, apply texture mapping technology to give the water a glossy effect; apply the fluid dynamics model to give the water cavity a dynamic effect;

[0135] Set a threshold for the area or volume of water holes to avoid filling real islands. After filling, verify whether the water holes are covered by plants.

[0136] Specifically, if Figure 6 and 7Comparison before and after filling water holes. The boundary rings of the holes are directly triangulated, and the newly generated triangles are smoothed to blend with the curvature of the surrounding water surface. The normals of the new vertices are adjusted to ensure consistency with the water surface. The elevation of the filled area is kept consistent with the surrounding water surface. To prevent overfilling, area / volume thresholds are set to avoid filling real islands. Image data is used to verify whether the holes are artifacts such as vegetation cover. The filled water area is beautified by adding surface reflections and water flow effects.

[0137] The S3 step achieves accurate definition and clear presentation of water areas in the three-dimensional model, providing solid and reliable data support for subsequent accurate analysis of water flow direction, precise prediction of rainy season inundation areas, and scientific assessment of inundation speed.

[0138] In this embodiment, step S4 includes:

[0139] Use lightweight tools to lightweight 3D real-scene models and optimize model structure and texture.

[0140] Specifically, the 3D real-world model is continuously iteratively optimized, with textures optimized. Previously blurry and distorted textures are now sharper and clearer, with richer details, making the entire model appear more realistic and vivid, meeting users' demands for high-quality visuals. Lightweight processing removes duplicate and redundant textures from the model, reducing the burden of texture loading and rendering, and improving overall performance. Developers can implement new interactive features and data analysis capabilities based on the 3D real-world model, reducing development costs and time.

[0141] Photos taken on sunny or rainy days, as well as different lenses, can cause uneven or distorted color in the aerial triangulation data, requiring software adjustment to correct the color. Considering the workload of color-matching the original photos using Photoshop, the team used MeshMaster in conjunction with DasViewer to color-match the model. This involves first using the color adjustment function in the model browser to color-match the model display, then exporting the model color-matching adjustment plan to MeshMaster. MeshMaster then batch-matches the entire model. This process of working backwards from the results to the color-matching plan ensures a more harmonious overall model effect.

[0142] Example 2

[0143] like Figure 6 As shown, a high-precision three-dimensional real scene model system generated based on a method for generating a high-precision three-dimensional real scene model includes:

[0144] Functional modules with 2D data, 3D data and directory tree;

[0145] Oblique photography loading effect diagram;

[0146] Disaster warning zoning module: color rendering of different areas, setting transparency and elevation data;

[0147] Risk point and hidden danger point module: Different colors represent different types of disaster points, which are plotted on the 3D real-life model. Click the icon to display the measurement point information.

[0148] Terrain elevation profile analysis module: select points to draw profile lines, display profile elevation information in a graphical format, and obtain coordinate information by moving within the oblique photography loading effect diagram;

[0149] Terrain Contour Analysis Module: There are two areas to choose from for drawing, namely rectangle and polygon. The contour lines can be controlled and displayed by adjusting the interval and line width sliders on the panel.

[0150] Slope and aspect analysis module: There are three drawing methods: rectangle, polygon and point. Different drawing methods have different processing times. The larger the drawing area, the longer the waiting time.

[0151] Cut and fill analysis module: draw the cut and fill area and calculate the volume of the cut area;

[0152] Flood analysis module: Draw the flooded area, control the altitude and flooding speed; slide to select the flooding height to understand the flooded area.

[0153] Based on a high-precision three-dimensional real-scene model system generated by a method for generating a high-precision three-dimensional real-scene model, the above-mentioned functions or modules are developed to enable dynamic monitoring and early warning of terrain changes caused by rainfall, thereby achieving accurate prediction and timely warning of the scope of flooding.

[0154] A method for generating high-precision 3D real-world models begins by capturing images of the target area and then constructing a 3D real-world model. During the model construction process, key processing is performed on the water area to effectively fill in voids and ensure model integrity. This high-precision 3D real-world model system can enhance dynamic monitoring and early warning capabilities for rainfall-induced surface changes, enabling accurate prediction and timely warning of flooding extent, providing solid and reliable technical support and a scientific basis for decision-making in disaster prevention and mitigation efforts.

[0155] It should also be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0156] The above description of the disclosed embodiments will enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for generating a high-precision three-dimensional real scene model, characterized in that the steps include: Data collection: Deploy image control points on the surface of the area; take aerial photos of the entire surface of the area to obtain several images; Constructing a basic tilted real-scene 3D model: Perform aerial triangulation based on the control point data, image data, and camera parameters to construct a basic tilted real-scene 3D model; Forming a 3D real scene model: Editing the water cavity based on the basic tilted real scene 3D model to form a 3D real scene model; specifically including: The Canny or Sobel algorithm is used to extract the boundaries of the waterside of the basic tilted real-life 3D model; semantic segmentation is performed on the boundaries to identify water areas, suspended areas below the ground, isolated water bodies with an area or volume below a threshold, and depressions as potential water areas; In the basic tilted real-scene 3D model, select the water area and outline the boundary of the water area; select the central island on the water surface and outline the boundary of the central island; select the isolated water body whose area or volume is lower than the threshold and outline the boundary of the isolated water body; select the depression area and outline the boundary of the depression; Based on the range lines formed by the boundary and the range line types, the water holes are reconstructed or filled until the image of the 3D reality model is complete.

2. The method for generating a high-precision three-dimensional real scene model according to claim 1, characterized in that: The reconstruction or filling of water holes includes: Constrain the elevation of the water cavity to be consistent with the surrounding water surface, apply texture mapping technology to give the water a glossy effect; apply the fluid dynamics model to give the water cavity a dynamic effect; Set a threshold for the area or volume of water holes to avoid filling real islands. After filling, verify whether the water holes are covered by plants.

3. The method for generating a high-precision three-dimensional real scene model according to claim 1, wherein: The step of arranging image control points on the surface of the area includes: Deploy image control points: Deploy one image control point every 5 kilometers on the surface of the area; Checkpoints: One checkpoint will be set up every 8 kilometers on the surface of the area; Measuring image control points: Obtain the planar position of image control points based on PTK measurement or total station traverse measurement; obtain the elevation position of image control points based on geometric leveling, total station trigonometric height measurement, or RTK height measurement combined with a refined regional quasi-geoid model; Combining the plane position and elevation position of each image control point, a combined map of the imaging control point distribution is drawn.

4. The method for generating a high-precision three-dimensional real scene model according to claim 2, wherein: The step of arranging image control points on the surface of the area further includes: Each image control point shall be observed for at least 3 rounds, and the difference of the coordinate components of the plane position shall be less than 2cm, and the difference of the coordinate components of the elevation position shall be less than 3cm. If it exceeds the range, the number of measurements shall be increased; each detection point shall be observed for at least 1 round, and the difference of the coordinate components of the detection point shall be less than 5cm. If it exceeds the range, image control points shall be added at both ends of the detection point and measured; the median of each image control point shall be taken as the final observation result.

5. The method for generating a high-precision three-dimensional real scene model according to claim 1, wherein: The steps of obtaining a plurality of images by aerial photography of the complete coverage of the surface in the area include: Set the flight difference mode: set to post-processing difference mode; Plan flight routes: Develop flight routes based on regional terrain, environmental restrictions, flight constraints, and weather changes. The flight routes include the locational relationship between the departure point, en route points, and destination point, flight altitude and speed, and the time period required to reach the destination. There should be navigation edges between flight routes, with a heading overlap of 53% to 65%, with a minimum of no less than 53%. The relative heading overlap should be 51% to 65%, and the heading overlap of adjacent image pairs should be no less than 58%. The route edges should extend beyond the detection point range and the mapping area. Flight data inspection: Perform quality inspection on flight data to ensure that the flight image data can effectively cover the area and ensure data quality; retake unqualified images; plan the retake route, and the two ends of the retake route should exceed a photographic baseline; for loopholes that do not affect the construction of the basic tilt real scene 3D model, retake only the loopholes; Data processing: Analyze the resolution, photographic scale, ground resolution and image flight relationship of aerial photographs; organize and analyze camera data, image control point collection, and aerial photograph index maps; and perform post-differential flight count calculation on UAV data.

6. The method for generating a high-precision three-dimensional real scene model according to claim 5, characterized in that: The steps for calculating the post-difference flight times of UAV data include: Calculate the location of base station data, mobile station data, base station coordinate information, antenna-camera relative position information, and base station instrument height information; Correction of differential offset: Calculate the offset value of the camera in the NEZ direction relative to the antenna, with the antenna as the center. Calculate the position data of each point in the real-time differential data according to the azimuth at the time of flight, and correct the offset value to the position information corresponding to each camera shooting point.

7. The method for generating a high-precision three-dimensional real scene model according to claim 1, wherein: The step of constructing a basic tilted real scene three-dimensional model comprises: Use automated data checking tools to ensure the data quality of acquired images meets requirements; Based on the original image and POS data, the project automatically connects the points and preliminarily determines the correspondence between different images. The project also uses the measured image control points to absolutely locate and orient the oblique image, correct the exterior orientation elements, restore the true position and attitude of each photo, complete the aerial triangulation, and generate the aerial triangulation report. Automatic creation of tilted real-scene 3D models based on aerial triangulation reports; Generate DSM and DOM results based on aerial triangulation reports.

8. The method for generating a high-precision three-dimensional real scene model according to claim 7, characterized in that: The automatic creation of a tilted real-scene 3D model based on an aerial triangulation report includes: retrieving the clearest images from each corresponding perspective for texture mapping; after the initial texture mapping is completed, applying a color balancing algorithm to the tiles with texture information and calculating an overall balance value; and applying the balance value to the subsequent texture mapping process of the 3D model.

9. The method for generating a high-precision three-dimensional real scene model according to claim 1, wherein: It also includes using lightweight tools to lightweight three-dimensional real-scene models and optimize model structure and texture.

10. A high-precision three-dimensional real scene model system generated by a method for generating a high-precision three-dimensional real scene model according to any one of claims 1 to 9, characterized in that: include: Functional modules with 2D data, 3D data and directory tree; Oblique photography loading effect diagram; Disaster warning zoning module: color rendering of different areas, setting transparency and elevation data; Risk point and hidden danger point module: Different colors represent different types of disaster points, which are plotted on the 3D real-life model. Click the icon to display the measurement point information. Terrain elevation profile analysis module: select points to draw profile lines, display profile elevation information in a graphical format, and obtain coordinate information by moving within the oblique photography loading effect diagram; Terrain Contour Analysis Module: There are two areas to choose from for drawing, namely rectangle and polygon. The contour lines can be controlled and displayed by adjusting the interval and line width sliders on the panel. Slope and aspect analysis module: There are three drawing methods: rectangle, polygon and point. Different drawing methods have different processing times. The larger the drawing area, the longer the waiting time. Cut and fill analysis module: draw the cut and fill area and calculate the volume of the cut area; Flood analysis module: Draw the flooded area, control the altitude and flooding speed; slide to select the flooding height to understand the flooded area.

Images