Modeling platform and modeling method based on rotary scanning device

Abstract

The invention relates to a modeling platform and method based on a rotary scanning device. The device comprises a base, a rotary holder module, a laser radar and a software processor. The modeling method comprises the following steps that: the laser radar scans a measured object at a set azimuth angle value to obtain point cloud data of the measured object; reading the point cloud data by the software processor, creating a plurality of point cloud contours, and constructing a triangulation network between adjacent contours; and combining the constructed triangulation networks to form a digital model of the measured object. Compared with the prior art, the method has the advantages that the original point cloud data collected by the laser radar is directly utilized, the model is constructed according to the azimuth angle information and the point cloud data, the point cloud data does not need to be filtered and corrected, and the modeling efficiency is improved; the azimuth angle and the connection characteristic of each point are known, and the modeling precision is high; a point cloud contour point missing repairing method is adopted, so that the problem of modeling defects caused by acquired data missing is solved; error correction is carried out before the laser radar scans the measured object, and the modeling precision is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of digital modeling, and in particular to a modeling platform and a modeling method based on a rotating scanning device. Background Art

[0002] LiDAR is a point set that can quickly obtain the surface shape of the object being measured and can accurately present the surface shape of the object being measured. It is an efficient data acquisition mode and is therefore used in many fields, such as: cultural relic surface reconstruction, unmanned driving, laser mapping, and so on.

[0003] There are many algorithms for constructing triangulated networks of three-dimensional discrete points, which can be divided into two categories: direct method and indirect method. The biggest problem with the direct method is that it is difficult to accurately judge the adjacency relationship between points when the point density is uneven or there is noise, which can easily lead to network construction errors, such as surface holes, facet overlap, and inconsistent normals. The results of the indirect method often depend on the selection of the K neighborhood size. Improper selection may lead to inaccurate network construction. In addition, parallel projection of the three-dimensional point set onto the plane can also easily lead to errors in judging the adjacency relationship between points.

[0004] After searching, application publication number CN116137056A discloses a facade sampling line generation method, device, equipment, and storage medium. Specifically, the following is disclosed: In the facade sampling line generation method, the point cloud data of the object, the preset upper top line and lower bottom line can be obtained. Based on the upper top line and lower bottom line, the sampling limit of the facade sampling line is determined, and the point cloud data within the sampling limit is triangulated to obtain multiple triangular planes. According to the preset sampling line interval value, the sampling points are determined in the multiple triangular planes, and according to the elevation of the sampling points, the elevation line is traced in the multiple triangular planes to obtain the facade sampling lines. However, this existing technology does not solve the problem of inaccurate meshing when the point density is uneven or there is noise.

[0005] In summary, how to design a modeling method that can accurately determine the connection relationship between points and accurately construct a network is a technical problem that needs to be solved. Summary of the Invention

[0006] The purpose of the present invention is to provide a modeling platform and a modeling method based on a rotating scanning device in order to overcome the defects of inaccurate network construction in the above-mentioned prior art.

[0007] The purpose of the present invention can be achieved by the following technical solutions:

[0008] According to one aspect of the present invention, a modeling platform based on a rotating scanning device is provided, which is placed in an object to be measured. The platform includes a base, a rotating pan-tilt platform module, a laser radar and a software processor. The laser radar is installed on the rotating pan-tilt platform module; a motor is provided in the base, and the motor output shaft is connected to the rotating pan-tilt platform module; the software processor is installed in the base, and the point cloud data of the object to be measured collected by the radar is used by the software processor to obtain multiple point cloud contours, and a triangulated mesh is constructed for the point cloud contours to obtain a digital model of the object to be measured.

[0009] According to another aspect of the present invention, a modeling method based on the modeling platform based on the rotating scanning device is provided, which specifically comprises the following steps:

[0010] Step S1: The laser radar scans the object to be measured at a set azimuth angle value to obtain point cloud data of the object to be measured;

[0011] Step S2: The software processor reads the point cloud data, creates multiple point cloud contours, and constructs triangulated networks between adjacent contours.

[0012] Step S3: merging the constructed multiple triangulated networks to form a digital model of the object to be measured.

[0013] As a preferred technical solution, in step S1, error correction is performed before the radar scans the object to be measured, specifically including the following steps:

[0014] Step S11: fix the calibration object, and scan the calibration object with a laser radar to obtain a first profile;

[0015] Step S12, rotating the laser radar by 180° to scan the calibration object to obtain a second profile;

[0016] Step S13: compare the first contour and the second contour, where the distance between the first contour and the second contour is the translation parameter of the laser radar.

[0017] As a preferred technical solution, the step S1 specifically includes the following steps:

[0018] Step S101, setting a determined azimuth value;

[0019] Step S102: The laser radar is rotated to either side of the horizontal plane by a fixed azimuth angle with the horizontal plane as 0° to scan the object;

[0020] Step S103: The laser radar is rotated again to the same side by a fixed azimuth angle to scan the object under test;

[0021] In step S104, after all scans on one side are completed, the radar returns to the horizontal plane and rotates to the other side by a fixed azimuth angle to scan the object under test, and step S103 is repeated until the scan is completed.

[0022] As a preferred technical solution, in step S104, when the laser radar rotates from 0° to 90° or -90° relative to the horizontal plane, it is determined that the laser radar has completed the scan on the side where the laser radar is located on the two horizontal sides.

[0023] As a preferred technical solution, the step S2 specifically includes the following steps:

[0024] Step S201, creating a point cloud contour list based on the point cloud data, and sorting the point cloud contours in the list;

[0025] Step S202, calculating the index of each point on each point cloud contour according to the azimuth angle and sorting them;

[0026] Step S203: connecting two adjacent points on a point cloud contour with a point on its adjacent point cloud contour to form a triangular mesh;

[0027] Step S204 : connecting the points on each contour in sequence to construct a triangulated network between two adjacent point cloud contours.

[0028] As a preferred technical solution, step S203 specifically includes the following steps:

[0029] Step S2021, define two adjacent point cloud contours as C i and C i+1 , i∈{2, 3, 4, …};

[0030] Step S2022, define C in sequence i The point on C i+1 The point on

[0031] Step S2023, click and Connected to each other to form a triangular mesh;

[0032] Step S2024, click and Connected to each other to form a triangular mesh;

[0033] Step S2025, repeat steps S2023 and S2024, and set C i and C i+1 Connect all the points on it.

[0034] As a preferred technical solution, in step S204, when a point on C1 or C2 is missing, the point on the adjacent contour corresponding to the missing point is skipped and does not participate in the triangular mesh construction; the skipped points are patched, and the patching equation is: in, For contour C i The set of points on.

[0035] As a preferred technical solution, in step S3, the method for merging multiple triangulated networks is the same as the method for constructing triangulated networks between adjacent contours in step S2.

[0036] As a preferred technical solution, the distance between the object to be measured and the laser radar is 1 to 50 meters.

[0037] Compared with the prior art, the present invention has the following beneficial effects:

[0038] 1) The present invention directly uses the original point cloud data collected by the laser radar to obtain the point cloud outline. According to the azimuth information, the points on the point cloud outline are triangulated to form a digital model. The original point cloud data does not need to be filtered or corrected, which improves the modeling efficiency. The azimuth and connection characteristics of each point are known, and the modeling accuracy is high.

[0039] 2) The present invention adopts a method for repairing missing point cloud contour points, which solves the problem of modeling defects caused by missing data collected by lidar;

[0040] 3) The present invention performs error correction on the laser radar before scanning the object to be measured, so that the constructed model of the object to be measured is closer to the real shape, further improving the modeling accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a structural schematic diagram of a modeling platform based on a rotating scanning device according to the present invention;

[0042] Figure 2 This is a flow chart of a modeling method of a modeling platform based on a rotating scanning device according to the present invention;

[0043] Figure 3 A first contour schematic diagram is obtained for radar error correction of the present invention;

[0044] Figure 4 A second contour schematic diagram is obtained for radar error correction of the present invention;

[0045] Figure 5 This is a schematic diagram of the first working position of the rotating pan-tilt module and laser radar combination of the present invention;

[0046] Figure 6This is a schematic diagram of the second working position of the rotating pan-tilt module and laser radar combination of the present invention;

[0047] Figure 7 This is a schematic diagram of point cloud contour sorting according to the present invention;

[0048] Figure 8 This is a schematic diagram of constructing a triangulated network of two adjacent point cloud contours according to the present invention;

[0049] Figure 9 This is a schematic diagram of merging multiple triangulated networks constructed in the present invention;

[0050] Figure 10 This is the point cloud rendering of the carton before calibration of the present invention;

[0051] Figure 11 This is the point cloud effect diagram of the carton after calibration of the present invention;

[0052] Figure 12 This is a comparison diagram between the Poisson network construction method of Example 3 of the present invention and the present invention;

[0053] Figure 13 This is a graph showing the change in the carton outline network size according to Example 3 of the present invention;

[0054] Figure 14 This is a carton size error variation curve diagram of Example 3 of the present invention;

[0055] Figure 15 This is a comparison diagram between the Poisson network construction method of Example 4 of the present invention and the present invention;

[0056] Figure 16 This is a curve diagram of the room outline network size change in Example 4 of the present invention;

[0057] Figure 17 This is a curve diagram of room size error variation in Example 4 of the present invention;

[0058] The numbers in the figure show:

[0059] 1. Base, 2. Pan-tilt module, 3. LiDAR, 30. Radar motor, 31. Transmitter, 32. Receiver, 4. Software processor, 5. Motor, 6. Calibration board. DETAILED DESCRIPTION

[0060] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0061] Example 1

[0062] like Figure 1 As shown, the present invention provides a modeling platform based on a rotating scanning device. This platform fully utilizes the initial information acquired during radar acquisition, such as the point cloud of the scan profile and azimuth angle, to achieve direct meshing without feature recognition. The platform includes a base 1, a rotating pan-tilt module 2, a laser radar 3, and a software processor 4.

[0063] A motor 5 is provided in the base 1 , and an output shaft of the motor 5 is connected to the rotating pan-tilt platform module 2 to drive the rotating pan-tilt platform to rotate. Specifically, the motor 5 used is a stepping motor.

[0064] The laser radar 3 is mounted on the rotating platform module 2, which drives the laser radar 3 to rotate. The laser radar 3 includes a radar motor 30, a transmitter 31, and a receiver 32. The radar motor 30 drives the transmitter 31 and the receiver 32 to rotate.

[0065] The software processor 4 is installed in the base 1 and includes a data processing module, a data modeling module and a data analysis module.

[0066] The device is fixed in position, powered by a battery, and transmits data wirelessly. When working, it controls the rotating pan-tilt module 2 to rotate horizontally to scan the object to be measured, and the laser radar 3 collects point cloud contour data. After the collection is completed, the collected data is processed and networked by the software processor 4 to obtain a digital model of the object to be measured.

[0067] Example 2

[0068] like Figure 2 As shown, the present invention provides a modeling method for a modeling platform based on a rotating scanning device. This method utilizes the original point cloud data collected by the laser radar 3 and does not require filtering or correction. This method utilizes the contour information formed by the rotating scanning of the single-line laser radar 3 and the azimuth information of each point in the contour to construct a network according to adjacent contours, thereby improving the speed and robustness of the network construction and being particularly suitable for certain specific application scenarios. The contours of the collected point cloud data are sorted, and then the point cloud on each contour is sorted using the azimuth information. The points on adjacent contours are networked in sequence. If there is a blank point on the azimuth, the corresponding points of the adjacent contours are skipped from the network. Finally, the points on the adjacent contours are networked in sequence according to the azimuth order, and the broken surfaces caused by the blank points are repaired to form the final triangular mesh. This method is composed of adjacent triangles, and the vertices of each triangle are on the contour. In this way, the azimuth and connection characteristics of each vertex can be obtained, which improves the efficiency of the network construction.

[0069] The modeling method is further explained below:

[0070] like Figure 3 and Figure 4As shown, LiDAR 3 requires error correction before scanning. This method uses an artificial stereo target for radar error correction. The calibration object is fixed and the radar is powered on and reset to zero. The initial profile is obtained by scanning the calibration object at its initial position. The radar is rotated 180° and the resulting profiles are compared. The distance between the two profiles is the radar's translation parameter in the X direction.

[0071] like Figure 5 and Figure 6 As shown in the figure, the laser radar 3 is fixed on the pan-tilt axis, and the pan-tilt axis is driven by the motor 5 to rotate horizontally to change the scanning direction of the laser radar 3. The coordinate system of the laser radar 3 is O-XYZ. As the axis rotates, the coordinate system O-XYZ will rotate accordingly, resulting in each frame of point cloud collected by the laser radar 3 not being in the same coordinate system. Therefore, each frame of point cloud needs to be projected to the pan-tilt axis coordinate system O L -X L Y L Z L In the example, due to the installation error, the laser radar 3 coordinate system is not aligned with the gimbal axis coordinate system, so the rotation matrix R1 is obtained by axis calibration; since the distance between the center of the laser radar 3 and the axis center is known, the coordinate system O-XYZ of each frame point cloud and the gimbal axis coordinate system O can be obtained. L -X L Y L Z L The translation vector between [0Δ y Δ z ] T Secondly, the coordinate system O-XYZ of the point cloud and the pan-tilt axis coordinate system O are obtained by rotating the motor 5 by 180°. L -X L Y L Z L Finally, the three-dimensional point (x'y'z') in the point cloud is projected into the gimbal axis coordinate system through equation (1).

[0072]

[0073] like Figure 7 As shown, the laser radar 3 is used to rotate and scan and collect the output as contour lines, generating a two-dimensional contour set D = {C1, C2, C3, ..., C i}, (i=1,2,3,…,n), each contour contains the azimuth v of each point j Information, sort the point set on the contour according to the azimuth; finally, the position relationship equation of each point on the contour in space is as follows:

[0074] P=(x i ,yi ,z i ,v j ),(i=1,2,3,…,n,j=1,2,3,…,C i ) (2)

[0075] Where x i ,y i ,z i are the position coordinates of the point in space, v j Expressed as the azimuth angle of the point on the i-th contour.

[0076] like Figure 8 As shown, in the process of meshing, first let the point set P on the adjacent single contour participate in the meshing, and then merge the two contours that have constructed triangulations. For example, according to the current contour C1 and vertex index Get the vertex index of the same orientation in the adjacent contour C2 Set the current contour vertex Next point with this contour Vertices of adjacent contours Construct a triangular mesh The vertex immediately following the adjacent contour The second point of the contour At the second point of the contour C1 Constructing a triangular mesh The final triangular mesh set obtained by this method is as follows:

[0077]

[0078] like Figure 9 As shown in the figure, during the scanning process, the contour at a certain azimuth angle may lose points, thus forming an empty data. Therefore, in this case, the adjacent contours at the azimuth angle of the missing point are not included in the network construction and the sequence point is skipped. For example, the index of the C1 contour point cloud set is The index of the C2 contour point cloud set with missing points P6 and P7 is Take a k value, indicating and The intersection of The result is k = {1, 2, 3, 4, 5, 8, ..., n}. The points of the k sets of C1 and C2 are added to the network, and the points P6 and P7 are skipped. The result is The contour set D={C1,C2,C3,…,C n The adjacent contours are meshed together to form the result: However, if we skip points P6 and P7, a broken surface will be formed when constructing the C3 and C2 contours. Therefore, the broken surface caused by the empty points should be repaired during the construction of the mesh. The repair equation is: The missing points of the two contours are connected and filled, and the broken surfaces are repaired. The contour lists are spliced ​​in sequence to form the final triangulated network. The final triangulated mesh set is as shown in formula (4):

[0079]

[0080] This meshing process can effectively avoid triangle intersections and reduce overlapping errors and broken surface errors of surface triangles during surface reconstruction.

[0081] Example 3

[0082] The present invention provides a modeling method for a modeling platform based on a rotating scanning device (since this method creates a point cloud contour and constructs a triangulated network between adjacent contours, this method is hereinafter referred to as a contour network method). The program running environment is an Intel (R) 4.Core (TM) i7-9750H CPU @ 2.60GHz, and a Windows operating system with 8GB of running memory is used as the experimental platform. When collecting point cloud data, a KR SN50 laser radar 3 is used to obtain point cloud data. The experimental conditions use a laser radar 3 to rotate and scan spaces of different sizes to obtain point cloud data and save them for experiments. The angle threshold is set to 180° and the number of scanning circles of the single-line laser radar 3 is 300. The specific data are shown in Table 1:

[0083] Table 1 Overview of experimental instruments

[0084]

[0085] In this embodiment, a cardboard box is selected as the object to be measured. After the laser radar 3 is calibrated, the calibration effect is verified by the cardboard box. The laser radar 3 determines the position through rotation scanning, and its position is relatively evenly and discretely distributed within the measurement range of the laser radar 3's measurement distance (1 to 50 m) and azimuth angle (±180°).

[0086] like Figure 10 and Figure 11 As shown in Figure 3, the calibrated data is closer to the true shape and effectively compensates for the impact of errors on the experimental results.

[0087] After calibration, the laser radar 3 rotates and scans the carton to form point cloud data. The obtained point cloud data is loaded into the Cloud Compare software and the Poisson algorithm is used to construct the network to compare with the network constructed by this method. Figure 12 (a) is the cardboard point cloud after normal vector and downsampling, 12 (c) is the original point cloud after contour network construction, Figure 12 Figures (b) and (d) show the results of Poisson meshing and contour meshing, respectively. The Poisson algorithm first calculates the normal vectors of all points when constructing a mesh, then downsamples them using an octree. After downsampling, the mesh is reconstructed. This method meshes the point clouds on adjacent contours sequentially by azimuth, skipping contours with no points at those azimuths. This results in a triangular mesh model that clearly shows the connection between each point. The length, width, and height measurements of the carton's inner wall are shown in Table 2.

[0088] Table 2 Carton length, width and height

[0089] Length (mm) Width (mm) Height(mm) Time(s) Carton inner wall size 685 505 750 \ Poisson grid size 689.03 504.35 779.89 22.88 Contour grid size 687.02 504.59 764.72 3.69

[0090] The laser rangefinder was used to manually measure the length dimensions of the research objects multiple times. The average values ​​of the measured dimensions were taken and used to manually calculate the volume values ​​of the two research objects, which was convenient for comparison with the results obtained by this method. The accuracy of the laser rangefinder was ±(0.25mm+5×10 -5 D), where D represents the measured distance. The experimental length is the length, width, and height of the carton after data processing and meshing, as measured using software measurement tools. Table 2 shows that this method achieves higher meshing accuracy than the Poisson algorithm. In terms of speed, the Poisson algorithm first calculates normal vectors and then downsamples before meshing, resulting in a complex algorithm flow and a meshing time of 22.88 seconds. This method, however, processes directly on the initial point cloud, achieving a meshing time of 3.69 seconds, significantly improving meshing speed.

[0091] The experimental data after multiple experiments are shown in Table 3.

[0092] Table 3 Carton test results

[0093]

[0094]

[0095] The experimental results show that the maximum dimensions of the length, width and height of the carton are 690.25, 508.76 and 766.41 mm; the minimum dimensions of the length, width and height of the carton are 680.52, 502.38 and 753.03 mm, and the range of the length, width and height of the carton is 9.73, 6.38 and 13.38 mm, respectively. The relative error with the actual size of the carton is as follows: Figure 13 The average relative error of length is 2.72mm, the average relative error of width is 1.88mm, and the average relative deviation of height is 9.06mm. Figure 14The model's prediction accuracy for height is slightly lower than that for length and width; the average relative error of the model's predictions for length, width, and height is less than 10 mm.

[0096] Example 4

[0097] The present invention provides a modeling method for a modeling platform based on a rotating scanning device (since this method creates a point cloud contour and constructs a triangulated network between adjacent contours, this method is hereinafter referred to as a contour network method). The program running environment is an Intel (R) 4.Core (TM) i7-9750H CPU @ 2.60GHz, and a Windows operating system with 8GB of running memory is used as an experimental platform. When collecting point cloud data, a KR SN50 laser radar 3 is used to obtain point cloud data. The experimental conditions use a laser radar 3 to rotate and scan spaces of different sizes to obtain point cloud data and save them for experiments. The angle threshold is set to 180° and the number of scanning circles of the single-line laser radar 3 is 300 circles. The specific data is the same as Table 1 in Example 3.

[0098] In this embodiment, a room is selected as the object to be measured, and the laser radar 3 is calibrated before scanning. Figure 15 (a) is the room point cloud after normal vector and downsampling. Figure 15 (c) is the original point cloud image of the contour network. Figure 15 Figures (b) and (d) show the results of Poisson meshing and contour meshing, respectively. As the space increases, the distribution of the three-dimensional point cloud data obtained by the rotational scanning of the single-line laser radar 3 becomes discrete, and the time required for both Poisson and contour meshing increases accordingly. In terms of accuracy, the Poisson meshing exhibits pit deformation as the space increases. This is because the Poisson algorithm includes the pit portion of the point cloud in the meshing, resulting in deformation after downsampling. However, this method does not process the point cloud data, so accuracy is not reduced. Compared with the method used in this embodiment, the contour meshing effect is better than the Poisson meshing effect. The room length, width, and height measurement results are shown in Table 4, where the actual length is the value obtained by measuring tools in the experimental scene, and the experimental length is the length, width, and height of the room measured using the software measurement tool after data processing and meshing. In terms of meshing speed, the Poisson meshing time is 1 minute and 37.70 seconds. The increase in point cloud data increases the normal vector calculation time and the meshing time will also slow down. The contour meshing time is 5.37 seconds. It can be seen that the Poisson meshing speed increases geometrically with the increase in point cloud data, while the contour meshing speed does not increase geometrically like the Poisson meshing speed with the increase in point cloud data.

[0099] Table 4 Room length, width and height

[0100] Length (mm) Width (mm) Height(mm) Time(s) Actual room size 3265 2745 2785 \ Poisson grid size 3364.67 2752.47 2766.84 1.37.70 Contour grid size 3279.62 2741.37 2772.13 5.37

[0101] The experimental data after multiple experiments are shown in Table 5.

[0102] Table 5 Room experiment results

[0103]

[0104] The experimental results show that the maximum dimensions of the length, width and height of the room are 3270.02, 2759.41 and 2790.48 mm; the minimum dimensions of the length, width and height of the carton are 3256.34, 2740.25 and 2770.95 mm, and the range of the length, width and height of the carton is 13.68, 19.16 and 19.53 mm, respectively. The relative error with the actual size of the room is as follows: Figure 16 The average relative error of length is 4.67mm, the average relative error of width is 5.81mm, and the average relative deviation of height is 4.41mm. Figure 17 The model's prediction accuracy for width is slightly lower than its prediction accuracy for length and height; the average relative error for length, width, and height predicted by the model is less than 10 mm. Based on the results from the carton in Example 3 and the room in this example, the average relative error is within 10 mm, fully verifying the accuracy and stability of this method.

[0105] Example 5

[0106] The present invention provides a modeling method for a modeling platform based on a rotating scanning device (since this method creates a point cloud contour and constructs a triangulated network between adjacent contours, this method is hereinafter referred to as a contour network method). The program running environment is an Intel (R) 4.Core (TM) i7-9750H CPU @ 2.60GHz, and a Windows operating system with 8GB of running memory is used as an experimental platform. When collecting point cloud data, a KR SN50 laser radar 3 is used to obtain point cloud data. The experimental conditions use a laser radar 3 to rotate and scan spaces of different sizes to obtain point cloud data and save them for experiments. The angle threshold is set to 180° and the number of scanning circles of the single-line laser radar 3 is 300 circles. The specific data is the same as Table 1 in Example 3.

[0107] This embodiment uses a corridor as the object to be measured, and scans the area after calibrating the laser radar 3. As the scanning radius increases, the point cloud becomes sparser during scanning, and the distance between two points increases. The Poisson algorithm does not completely eliminate these distant and sparse points after downsampling during meshing, and they also participate in the meshing. This can easily lead to errors between distant point clouds, resulting in inaccurate meshing and deformation. In terms of speed, the computer frequently crashes during Poisson meshing, while contour meshing does not.

[0108] According to Examples 3, 4, and 5, when the azimuth and contour information are known, this method is superior to the traditional Poisson meshing algorithm in terms of meshing accuracy and speed, and can still produce better meshing effects for distant and sparse point clouds.

[0109] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and such modifications or substitutions are intended to be within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be subject to the scope of protection of the claims.

Claims

1. A modeling platform based on a rotating scanning device, placed in the object to be measured, characterized in that: The platform comprises a base (1), a rotating platform module (2), a laser radar (3) and a software processor (4), wherein the laser radar (3) is mounted on the rotating platform module (2); a motor (5) is provided in the base (1), and an output shaft of the motor (5) is connected to the rotating platform module (2); the software processor (4) is mounted in the base (1), and point cloud data of the object to be measured collected by the radar is used to obtain a plurality of point cloud contours through the software processor (4), and a triangular mesh is constructed for the point cloud contours to obtain a digital model of the object to be measured.

2. A modeling method based on the modeling platform based on the rotating scanning device according to claim 1, characterized in that: The specific steps include: Step S1, the laser radar (3) scans the object to be measured at a set azimuth angle value to obtain point cloud data of the object to be measured; Step S2, the software processor (4) reads the point cloud data, creates a plurality of point cloud contours, and constructs a triangulated network between adjacent contours; Step S3: merging the constructed multiple triangulated networks to form a digital model of the object to be measured.

3. The modeling method according to claim 2, characterized in that In step S1, error correction is performed before the radar scans the object to be measured, which specifically includes the following steps: Step S11, fixing the calibration object, and scanning the calibration object with the laser radar (3) to obtain a first profile; Step S12, rotating the laser radar (3) by 180° to scan the calibration object to obtain a second profile; Step S13, comparing the first profile and the second profile, wherein the distance between the first profile and the second profile is a translation parameter of the laser radar (3).

4. The modeling method according to claim 2, characterized in that The step S1 specifically includes the following steps: Step S101, setting a determined azimuth value; Step S102, the laser radar (3) takes the horizontal plane as 0° and rotates to either side of the horizontal plane by a fixed azimuth angle value to scan the object to be measured; Step S103, the laser radar (3) rotates again to the same side at a fixed azimuth angle value to scan the object to be measured; In step S104, after all scans on one side are completed, the radar returns to the horizontal plane and rotates to the other side by a fixed azimuth angle to scan the object under test, and step S103 is repeated until the scan is completed.

5. The modeling method according to claim 4, characterized in that: In step S104, when the laser radar (3) rotates from 0° to 90° or -90° relative to the horizontal plane, it is determined that the laser radar (3) has completed scanning on the side where the laser radar (3) is located on the two horizontal sides.

6. The modeling method according to claim 2, characterized in that: The step S2 specifically includes the following steps: Step S201, creating a point cloud contour list based on the point cloud data, and sorting the point cloud contours in the list; Step S202, calculating the index of each point on each point cloud contour according to the azimuth angle and sorting them; Step S203: connecting two adjacent points on a point cloud contour with a point on its adjacent point cloud contour to form a triangular mesh; Step S204 : connecting the points on each contour in sequence to construct a triangulated network between two adjacent point cloud contours.

7. The modeling method according to claim 6, characterized in that: The step S203 specifically includes the following steps: Step S2021, define two adjacent point cloud contours as C i and C i+1 , i∈{2, 3, 4, …}; Step S2022, define C in sequence i The point on C i+1 The point on Step S2023, click and Connected to each other to form a triangular mesh; Step S2024, click and Connected to each other to form a triangular mesh; Step S2025, repeat steps S2023 and S2024, and set C i and C i+1 Connect all the points on it.

8. The modeling method according to claim 7, characterized in that: In step S204, when a point on C1 or C2 is missing, the points on the adjacent contour corresponding to the missing point are skipped and do not participate in the triangular mesh construction; the skipped points are patched, and the patching equation is: in, For contour C i The set of points on.

9. The modeling method according to claim 2, characterized in that: In step S3, the method for merging multiple triangulated networks is the same as the method for constructing triangulated networks between adjacent contours in step S2.

10. The modeling method according to claim 2, characterized in that: The distance between the object to be measured and the laser radar (3) is 1 to 50 m.

Images