A method and device for combining three-dimensional point cloud data of a large-span object

Abstract

The application discloses a large-span object three-dimensional point cloud data combination method and device, and the method is characterized in that: the front and rear partial point cloud data of a large-span object obtained by at least two laser radars in a measuring area are spliced, and the coordinate system conversion correction is performed by a reference body arranged in the measuring area in an early stage, so that a small part of coincident point cloud data in the point cloud data obtained by the two laser radars can be spliced, the point cloud collected by the double cloud tables is spliced, and the point cloud data of a super-large-size object is obtained.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-dimensional detection, in particular to a large-span object three-dimensional point cloud data combination method and device. BACKGROUND

[0002] With the development of science and technology and the wide application of computers and high-tech, digital stereophotogrammetry has gradually developed and matured, so radar positioning scanning devices also appear in the technology market. However, the traditional radar device can only scan two-dimensional icons. In recent years, with the continuous progress of science and technology, radar devices are also constantly improving. For example, laser radar devices can not only scan two-dimensional images, but also scan three-dimensional images. However, such laser radars that can scan a wide field of view and large objects are only limited to large occasions such as military and aviation.

[0003] However, there are many places where laser radars are needed at present, such as loading machines in factories, security checks, etc. However, the single scanning range of the laser radar device suitable for industrial places such as loading machines in the current market is limited. When the object to be scanned and measured is too large, the entire range of the object cannot be completely scanned. If multiple laser radars are used for segmented scanning, the point cloud data obtained by each laser radar must be spliced. The existing multi-point cloud data splicing mainly performs least squares fitting on the data of each point of the main point cloud. However, since the point cloud of the same part is very small and lacks features or angles, the fitting algorithm will fail, and the splicing of multiple point cloud data cannot be effectively completed. SUMMARY

[0004] The present application provides a large-span object three-dimensional point cloud data combination method to splice the front and rear partial point cloud data of a large-span object obtained by at least two laser radars installed in a measurement area, which comprises the following steps:

[0005] S1, obtaining first point cloud data containing a reference body in a first cloud head coordinate system by a first laser radar, and obtaining second point cloud data containing a reference body in the first cloud head coordinate system by a second laser radar;

[0006] S2, obtaining a first transformation matrix according to the installation position of the first laser radar, and converting the first point cloud data into third point cloud data in a first rough world coordinate system by the first transformation matrix;

[0007] obtaining a second transformation matrix according to the installation position of the second laser radar, and converting the second point cloud data into fourth point cloud data in a second rough world coordinate system by the second transformation matrix;

[0008] S3, converting the third point cloud data into two-dimensional images located in three coordinate planes respectively, obtaining a first angle offset matrix according to the reference body offset angle in the two-dimensional image, combining the first angle offset matrix with the first transformation matrix to form a third transformation matrix;

[0009] converting the fourth point cloud data into two-dimensional images located in three coordinate planes respectively, obtaining a second angle offset matrix according to the reference body offset angle in the two-dimensional image, combining the second angle offset matrix with the second transformation matrix to form a fourth transformation matrix;

[0010] S4, converting the third point cloud data into fifth point cloud data in the first accurate world coordinate system through the third transformation matrix; converting the fourth point cloud data into sixth point cloud data in the second accurate world coordinate system through the fourth transformation matrix; obtaining an offset matrix according to the position of the same reference body in the fifth point cloud data and the sixth point cloud data;

[0011] S5, after the point cloud data of the measured object obtained by the first laser radar is transformed through the third transformation matrix, the point cloud data is superimposed with the point cloud data obtained by the second laser radar after the point cloud data of the measured object is transformed through the fourth transformation matrix and the offset matrix to obtain the complete point cloud data of the measured object.

[0012] Preferably, the step S1 specifically comprises:

[0013] S11, obtaining first point cloud data containing first, second and third reference bodies in the gimbal coordinate system through the first laser radar; the first to fourth reference bodies are placed in sequence along the region central axis in the measurement region, wherein the second reference body and the third reference body are located in the measurement intersection region of the two laser radars, and the side of the second reference body and the side of the third reference body opposite to the side of the second reference body are located on the region central axis, the rear side of the second reference body and the front side of the third reference body are located on the vertical plane perpendicular to the region central axis, and the first and fourth reference bodies are located in the first laser radar measurement region and the second laser radar measurement region respectively, and the same side of the first and fourth reference bodies is located on the region central axis;

[0014] S12, obtaining second point cloud data containing second, third and fourth reference bodies in the gimbal coordinate system through the second laser radar.

[0015] Preferably, the step S3 further comprises:

[0016] converting the third point cloud data into a first two-dimensional image located on the XY coordinate plane, taking the Z-axis data of each point as the corresponding gray value of each point on the first two-dimensional image, identifying the connecting line of the first reference body close to the region central axis side and the intersection side of the second and third reference bodies in the first two-dimensional image, and calculating the included angle between the connecting line and the image edge as the Z-axis deflection angle of the third point cloud data;

[0017] The third point cloud data is converted into a second two-dimensional image located on the XZ coordinate plane, the Y-axis data of each point is taken as the corresponding gray value of each point on the second two-dimensional image, and the included angle between the ground and the image edge is recognized and obtained as the Y-axis deflection angle of the third point cloud data.

[0018] The third point cloud data is converted into a third two-dimensional image located on the YZ coordinate plane, the X-axis data of each point is taken as the corresponding gray value of each point on the third two-dimensional image, and the included angle between the ground and the image edge is recognized and obtained as the X-axis deflection angle of the third point cloud data.

[0019] The obtained axis deflection angles of the third point cloud data are combined into a first angle offset matrix, and the first angle offset matrix is combined with the first transformation matrix to form a third transformation matrix.

[0020] Preferably, the step S3 further comprises:

[0021] The fourth point cloud data is converted into a fourth two-dimensional image located on the XY coordinate plane, the Z-axis data of each point is taken as the corresponding gray value of each point on the fourth two-dimensional image, and the included angle between the connecting line of the intersection side of the second and third reference bodies and the axis side of the first reference body in the close area and the image edge is calculated and obtained as the Z-axis deflection angle of the fourth point cloud data.

[0022] The fourth point cloud data is converted into a fifth two-dimensional image located on the XZ coordinate plane, the Y-axis data of each point is taken as the corresponding gray value of each point on the fifth two-dimensional image, and the included angle between the second, third and fourth reference body bottom edges and the image edge is recognized and obtained as the Y-axis deflection angle of the fourth point cloud data.

[0023] The fourth point cloud data is converted into a sixth two-dimensional image located on the YZ coordinate plane, the X-axis data of each point is taken as the corresponding gray value of each point on the sixth two-dimensional image, and the included angle between the second, third and fourth reference body bottom edges and the image edge is recognized and obtained as the X-axis deflection angle of the fourth point cloud data.

[0024] The obtained axis deflection angles of the fourth point cloud data are combined into a second angle offset matrix, and the second angle offset matrix is combined with the second transformation matrix to form a fourth transformation matrix.

[0025] Preferably, the step S4 further comprises: obtaining a position offset of the second reference body or the third reference body in the fifth point cloud data and the sixth point cloud data, and obtaining an offset matrix according to the offset of each axis.

[0026] The application further discloses a three-dimensional detection system for obtaining three-dimensional data of a large-span object, comprising at least two laser radars installed in a measurement area and a controller in communication with the laser radars, wherein the laser radars are used to obtain partial point cloud data of front and rear parts of the large-span object respectively and send the data to the controller, and the controller is configured to:

[0027] obtain first point cloud data containing the reference body in a gimbal coordinate system through the first laser radar and obtain second point cloud data containing the reference body in the gimbal coordinate system through the second laser radar;

[0028] obtain a first transformation matrix according to the installation position of the first laser radar and convert the first point cloud data into third point cloud data in a first rough world coordinate system through the first transformation matrix, and obtain a second transformation matrix according to the installation position of the second laser radar and convert the second point cloud data into fourth point cloud data in a second rough world coordinate system through the second transformation matrix;

[0029] convert the third point cloud data into two-dimensional images in three coordinate planes respectively, obtain a first angle offset matrix according to the offset angle of the reference body in the two-dimensional images, combine the first angle offset matrix with the first transformation matrix to form a third transformation matrix, convert the fourth point cloud data into two-dimensional images in the three coordinate planes respectively, obtain a second angle offset matrix according to the offset angle of the reference body in the two-dimensional images, and combine the second angle offset matrix with the second transformation matrix to form a fourth transformation matrix;

[0030] convert the third point cloud data into fifth point cloud data in a first accurate world coordinate system through the third transformation matrix, convert the fourth point cloud data into sixth point cloud data in a second accurate world coordinate system through the fourth transformation matrix, and obtain an offset matrix according to the positions of the same reference body in the fifth point cloud data and the sixth point cloud data;

[0031] the point cloud data of the measured object obtained by the first laser radar is transformed through the third transformation matrix, and the point cloud data of the measured object obtained by the second laser radar is transformed through the fourth transformation matrix and the offset matrix, and then the transformed point cloud data is superimposed to obtain complete point cloud data of the measured object.

[0032] Preferably, the controller is configured to: acquire, by the first laser radar, first point cloud data containing the first, second and third reference bodies in the gimbal coordinate system; the first to fourth reference bodies are placed in sequence along the region central axis in the measurement region, wherein the second reference body and the third reference body are located in the measurement intersection region of the two laser radars, and the side of the second reference body and the side of the third reference body opposite to the side of the second reference body are located on the region central axis, the rear side of the second reference body and the front side of the third reference body are located on the vertical plane perpendicular to the region central axis, and the first and fourth reference bodies are located in the non-intersecting first laser radar measurement region and the second laser radar measurement region respectively, and the same side of the first and fourth reference bodies is located on the region central axis; and acquire, by the second laser radar, second point cloud data containing the second, third and fourth reference bodies in the gimbal coordinate system.

[0033] Preferably, the controller is configured to: convert the third point cloud data into a first two-dimensional image located on an XY coordinate plane, take the Z-axis data of each point as the corresponding gray value of each point on the first two-dimensional image, identify the line connecting the side of the first reference body close to the region central axis and the intersection side of the second and third reference bodies in the first two-dimensional image, and calculate the included angle between the line and the image edge as the Z-axis deflection angle of the third point cloud data.

[0034] convert the third point cloud data into a second two-dimensional image located on an XZ coordinate plane, take the Y-axis data of each point as the corresponding gray value of each point on the second two-dimensional image, and identify the included angle between the bottom edge of the first reference body, the bottom edge of the second reference body, the bottom edge of the third reference body, or the ground and the image edge on the second two-dimensional image as the Y-axis deflection angle of the third point cloud data.

[0035] convert the third point cloud data into a third two-dimensional image located on a YZ coordinate plane, take the X-axis data of each point as the corresponding gray value of each point on the third two-dimensional image, and identify the included angle between the bottom edge of the first reference body, the bottom edge of the second reference body, the bottom edge of the third reference body, or the ground and the image edge on the third two-dimensional image as the X-axis deflection angle of the third point cloud data.

[0036] combine the first angle offset matrix with the first transformation matrix to form a third transformation matrix.

[0037] Preferably, the controller is configured to:

[0038] The fourth point cloud data is converted into a fourth two-dimensional image located on an XY coordinate plane, Z-axis data of each point is taken as a corresponding gray value of each point on the fourth two-dimensional image, a line connecting a second reference body intersection side and a first reference body close area axis side in the fourth two-dimensional image is recognized and obtained, and an included angle between the line and an image edge is taken as a Z-axis deflection angle of the fourth point cloud data;

[0039] The fourth point cloud data is converted into a fifth two-dimensional image located on an XZ coordinate plane, Y-axis data of each point is taken as a corresponding gray value of each point on the fifth two-dimensional image, and an included angle between a second reference body bottom edge, a third reference body bottom edge, a fourth reference body bottom edge or a ground and an image edge on the fifth two-dimensional image is recognized and obtained as a Y-axis deflection angle of the fourth point cloud data;

[0040] The fourth point cloud data is converted into a sixth two-dimensional image located on a YZ coordinate plane, X-axis data of each point is taken as a corresponding gray value of each point on the sixth two-dimensional image, and an included angle between a second reference body bottom edge, a third reference body bottom edge, a fourth reference body bottom edge or a ground and an image edge on the sixth two-dimensional image is recognized and obtained as an X-axis deflection angle of the fourth point cloud data;

[0041] The obtained each-axis deflection angle of the fourth point cloud data is combined to form a second angle offset matrix, and the second angle offset matrix is combined with the second transformation matrix to form a fourth transformation matrix.

[0042] The application further discloses a large-span object three-dimensional point cloud data combination device for splicing front and rear partial point cloud data of a large-span object obtained by two laser radars installed in a measurement area.

[0043] The large-span object three-dimensional point cloud data combination method and device disclosed by the application can splice front and rear partial point cloud data of a large-span object obtained by at least two laser radars installed in a measurement area, and can splice point cloud data collected by two gimbals by means of a small amount of coincident point cloud data in the point cloud data obtained by the two laser radars, so that point cloud data of a super-large-size object can be obtained.

[0044] Additional aspects and advantages of the application will be made apparent by the following description. BRIEF DESCRIPTION OF DRAWINGS

[0045] The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application. In the drawings:

[0046] Figure 1 A flowchart of a method for combining three-dimensional point cloud data of a large-span object according to an embodiment of the present application.

[0047] Figure 2 And 7 A schematic diagram of a reference body for a measurement region according to an embodiment of the present application.

[0048] Figures 3-5 A schematic diagram of first point cloud data and second point cloud data according to an embodiment of the present application.

[0049] Figure 6 A schematic diagram of a specific flow of step S1 according to an embodiment of the present application.

[0050] Figure 8 A schematic diagram of a specific flow of step S3 according to an embodiment of the present application.

[0051] Figure 9 A schematic diagram of first point cloud data according to an embodiment of the present application.

[0052] Figure 10 A schematic diagram of a first two-dimensional image according to an embodiment of the present application.

[0053] Figure 11 A schematic diagram of a second two-dimensional image according to an embodiment of the present application.

[0054] Figure 12 A schematic diagram of a third two-dimensional image according to an embodiment of the present application.

[0055] Figure 13 A schematic diagram of a specific flow of step S3 according to another embodiment of the present application.

[0056] Figure 14 A schematic diagram of converted first point cloud data according to an embodiment of the present application.

[0057] Figure 15 A schematic diagram of converted second point cloud data according to an embodiment of the present application.

[0058] Figure 16 A schematic diagram of spliced and combined point cloud according to an embodiment of the present application.

[0059] Figure 17 A schematic diagram of spliced and combined three-dimensional point cloud according to an embodiment of the present application. DETAILED DESCRIPTION

[0060] In order to make the objects, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions of the embodiments of the present application will be clearly and completely described below with reference to the drawings of the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments of the present application. Based on the described embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present application.

[0061] In the present application, unless specifically defined and limited otherwise, the terms "mounting", "connection", "connecting", "fixed", and the like should be understood broadly, for example, can be fixed connection, can also be detachable connection, or integrally connected; can be mechanical connection, can also be electrical connection; can be directly connected, can also be indirectly connected through an intermediate medium, can be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present application can be understood according to the specific circumstances.

[0062] In the present application, unless specifically defined and limited otherwise, the first feature "on" or "under" the second feature can include that the first and second features are in direct contact, or that the first and second features are not in direct contact but are in contact through another feature between them. Moreover, the first feature "on", "above" and "over" the second feature includes that the first feature is directly above and obliquely above the second feature, or only means that the horizontal height of the first feature is higher than that of the second feature. The first feature "under", "below" and "under" the second feature includes that the first feature is directly below and obliquely below the second feature, or only means that the horizontal height of the first feature is less than that of the second feature.

[0063] Unless otherwise defined, the technical terms or scientific terms used herein should be understood as the usual meanings understood by those of ordinary skill in the art to which the present application belongs. The "first", "second" and similar words used in the patent application description and claims of the present application do not represent any order, quantity or importance, but are only used to distinguish different components. Similarly, "one" or "a" and similar words do not represent a quantity limit, but represent the existence of at least one.

[0064] Due to the limited scanning range of a single gimbal mounted laser radar, it is impossible to achieve complete scanning of a large-span object, so it can only be achieved by segmented scanning of multiple laser radars. The purpose of double or even multiple gimbals is to expand the scanning range of the point cloud, and in order to expand the range as much as possible, the overlapping part between the gimbal point clouds will be very small, and the main part scanned by different gimbals is different, so the mutual position relationship between the gimbals must be calculated from the very small overlapping point cloud part, or the position relationship between the gimbals is accurately known during installation, but the ultra-high precision mutual position and angle relationship cannot be guaranteed in the actual installation environment. The large-span object three-dimensional point cloud data combination method disclosed in the present application can realize point cloud splicing according to very small overlapping point clouds, and at the same time can convert the point cloud coordinate system to the device coordinate system with high precision, ensuring that the measurement positioning can be directly used. The large-span object three-dimensional point cloud data combination method disclosed in the present embodiment is used to splice the front and rear partial point cloud data of a large-span object obtained by two laser radars installed in a measurement area. In the following embodiments, a cement plate loading machine working site will be taken as an example to detect the three-dimensional size of a large length of vehicle to be loaded under the loading machine, so as to obtain point cloud data of the vehicle, especially the vehicle compartment, and finally obtain size information of the vehicle compartment. Two or more gimbals with laser radars can be installed in front of and behind the vehicle parking area above the to-be-loaded area, of course, the large-span object three-dimensional point cloud data combination method can also be applied to any scene where three-dimensional point cloud data of a large-span object is obtained. As shown in the accompanying Figure 1 The large-span object three-dimensional point cloud data combination method disclosed in the present embodiment specifically includes the following contents.

[0065] Step S1, obtaining first point cloud data containing a reference body in the first gimbal coordinate system by a first laser radar, and obtaining second point cloud data containing a reference body in the second gimbal coordinate system by a second laser radar.

[0066] Wherein the point cloud is a data set, each point in the data set represents a set of X, Y, Z geometric coordinate values, and the collection of data points in space represents a 3D shape or object. The gimbal is an institution for producing point clouds by rotating radars. Hereinafter, the world coordinate system w is introduced, and the world coordinate system has a certain number of translation relationships with the known device coordinate system, wherein the horizontal ground is the xy plane. The x-axis direction is the direction of the device axis pointing to the tail end of the device running, the y-axis direction is the direction of rotating the x-axis 90° clockwise in the view of the ground, and the z-axis direction is the direction perpendicular to the xy plane downward. The origin position can be specifically set and adjusted according to the situation, such as the accompanying Figure 2 and 3As shown, in the embodiment, the x-axis zero point can be the position where the front and rear running stroke distance of the device is equal to zero, the y-axis zero point is the position where the middle axis of the device moves 3 meters in the y negative direction, and the z-axis zero point is the position where the xy plane moves 6 meters in the z negative direction.

[0067] Wherein the coordinate system direction of the first gimbal coordinate system c1 is determined according to the current installation direction, as shown in the attached Figure 4 As shown, the x-axis direction is approximately the same as the y-axis direction of the world coordinate system but has an installation deviation, the y-axis direction is approximately the same as the x-axis direction of the world coordinate system but has an installation deviation, and the z-axis direction is approximately the same as the z-axis direction of the world coordinate system but has an installation deviation, wherein the origin position is the actual installation position of the first gimbal.

[0068] Wherein the coordinate system direction of the second gimbal coordinate system c2 is determined according to the current installation direction, as shown in the attached Figure 5 As shown, the x-axis direction is approximately the same as the y-axis direction of the world coordinate system but has an installation deviation, the y-axis direction is approximately the same as the x-axis direction of the world coordinate system but has an installation deviation, and the z-axis direction is approximately the same as the z-axis direction of the world coordinate system but has an installation deviation, wherein the origin position is the actual installation position of the first gimbal.

[0069] In the embodiment, as shown in the attached Figure 6 As shown, the step S1 can further include the following contents.

[0070] Step S11, acquiring, by the first laser radar, first point cloud data containing the first, second and third reference bodies in the gimbal coordinate system; the first to fourth reference bodies are placed in the measuring area along the area center axis direction in sequence, wherein the second reference body and the third reference body are located in the measuring intersection area of the two laser radars, and the side of the second reference body and the side of the third reference body opposite to the side of the second reference body are located on the area center axis, the rear side of the second reference body and the front side of the third reference body are located on the vertical plane perpendicular to the area center axis, and the first and fourth reference bodies are located in the non-intersecting first laser radar measuring area and second laser radar measuring area respectively, and the same side of the first and fourth reference bodies is located on the area center axis.

[0071] Step S12, acquiring, by the second laser radar, second point cloud data containing the second, third and fourth reference bodies in the gimbal coordinate system.

[0072] Specifically, the embodiment places multiple reference bodies in the measuring area to assist in the conversion of the coordinate system, that is, rotates and translates multiple pieces of point cloud data to a unified coordinate system so that they can form complete environmental point cloud data. As shown in the attached Figure 2 and 7As shown, in the embodiment, four reference bodies are accurately placed in the measurement area, the reference bodies are preferably cuboid in shape, but can also be other shapes or quantities, and the processing process is similar. In the device in the embodiment, the device axis is the central axis of the loading machine, that is, the central axis of the movement area thereof, the device axis reference line of the ground of the measurement area can be obtained by placing a laser ruler or the like, and then the placement of the reference bodies is performed according to the reference line. The first reference body 1, the second reference body 2, the third reference body 3 and the fourth reference body 4 are arranged in sequence along the central axis of the area from the first holder to the second holder. The first reference body and the fourth reference body are respectively located in the first laser radar measurement area and the second laser radar measurement area which do not intersect, and the same side of the first reference body and the fourth reference body is placed on the central axis of the area. The second reference body and the third reference body are located in the measurement intersection area of the two laser radars, and the second reference body and the third reference body are arranged on both sides of the central axis of the area, and the side of the second reference body opposite to the side of the third reference body is placed on the central axis of the area. The rear side of the second reference body and the front side of the third reference body are respectively located on a vertical plane perpendicular to the central axis of the area. At the same time, the rear edge of the first reference body can be placed at a position where the front-rear running stroke distance of the device is equal to zero, that is, the x-axis zero point of the world coordinate system W.

[0073] In step S2, a first transformation matrix is obtained according to the installation position of the first laser radar, and the first point cloud data is converted into third point cloud data in a first rough world coordinate system through the first transformation matrix; a second transformation matrix is obtained according to the installation position of the second laser radar, and the second point cloud data is converted into fourth point cloud data in a second rough world coordinate system through the second transformation matrix.

[0074] The first transformation matrix is obtained according to the installation position of the first laser radar. According to the installation position of the first holder on which the first laser radar is installed, the first holder coordinate system c1 is rotated by 90 degrees around the z-axis, moved 6m in the positive direction of the x-axis, and moved 2m in the positive direction of the y-axis through the first transformation matrix After the change, a first rough world coordinate system w' is obtained, the xyz axes of the first rough world coordinate system and the first accurate world coordinate system are substantially in the same direction, but there is still a small included angle caused by installation error. That is, the first point cloud data is rotated by 90 degrees around the z-axis, moved 6m in the positive direction of the x-axis, and moved 2m in the positive direction of the y-axis, and the first point cloud data is converted into third point cloud data.

[0075] If the first holder coordinate system c1 is to be converted into a first accurate world coordinate system which is highly accurately orthogonal to the world coordinate system, c1 needs to be rotated by 90 degrees around the z-axis, moved 6m in the positive direction of the x-axis, and moved 2m in the positive direction of the y-axis through After the change, a coordinate system w" which is highly accurately orthogonal to w is obtained, and the included angles of the xyz axes of the first rough world coordinate system and the first accurate world coordinate system are α, β and γ.

[0076] Similarly, a second transformation matrix is obtained according to the second laser radar installation position, and the second point cloud data is converted into fourth point cloud data in a second rough world coordinate system through the second transformation matrix, which is not repeated here.

[0077] In step S3, the third point cloud data is converted into two-dimensional images in three coordinate planes respectively, a first angle offset matrix is obtained according to the reference body offset angle in the two-dimensional images, the first angle offset matrix is combined with the first transformation matrix to form a third transformation matrix, and the fourth point cloud data is converted into two-dimensional images in three coordinate planes respectively, a second angle offset matrix is obtained according to the reference body offset angle in the two-dimensional images, and the second angle offset matrix is combined with the second transformation matrix to form a fourth transformation matrix. In this embodiment, as shown in the attached Figure 8 , step S3 can specifically include the following contents.

[0078] In step 101, the third point cloud data is converted into a first two-dimensional image in the XY coordinate plane, the Z-axis data of each point is taken as the corresponding gray value of each point in the first two-dimensional image, the connecting line of the first reference body close area on the axis side and the intersection side of the second and third reference bodies in the first two-dimensional image is identified and obtained, and the included angle between the connecting line and the image edge is calculated as the Z-axis deflection angle of the third point cloud data.

[0079] Specifically, the projection of each point of the point cloud in the plane xoy constitutes the position of each pixel in the two-dimensional image, and the gray value of the pixel can be the x value, y value or z value of the point, or the proportional value. For example, the 3D point cloud point n (x n ,y n ,z n ) is converted into a 2D two-dimensional image, and z n is defined as the image gray value, that is, the 2D image has a point image with x=x n , y=y n , and z n as the gray value at the position.

[0080] As described in the foregoing step S2, according to the installation position of the first holder, a rough conversion relationship Pose front (6, 2, 0, 0, 0, 90) between the first holder coordinate system c1 and the world coordinate system w, that is, the first transformation matrix, can be obtained. The first holder coordinate system c1 is changed through the first transformation matrix to obtain the first rough world coordinate system w', and the xyz axes of the first rough world coordinate system w' and the world coordinate system w are basically in the same direction, but there is still a small included angle. The included angle values between the xyz axes of the two coordinate systems w' and w are calculated respectively, and the following transformation can be performed:

[0081] Through Pose over (6, 2, 0, 0, 0, 90) transformation calculated ground is the rotation angle around the z-axis of the w coordinate system.

[0082] Through Pose side (6, 0, 3, 270, 0, 90) transformation calculated ground is the rotation angle around the y-axis of the w coordinate system.

[0083] Through Pose back (6, 0, 10, 270, 0, 180) transformation calculated ground is the rotation angle around the x-axis of the w coordinate system.

[0084] First calculate the angle between w' and the x-axis of w, c1 point cloud through Change. Attached Figure 9 is the point cloud of c1 through Pose over (6, 2, 0, 0, 0, 90), attached Figure 10 is the 2d image converted from the 3D point cloud, where the arrowed horizontal line is the connecting line between the side of the first reference body close to the axis and the intersection side of the second and third reference bodies. Through the 2d image algorithm, the angle between the arrowed horizontal line and the horizontal axis in the above figure can be calculated, that is, the rotation angle between w' and the z-axis of w is λ', and the horizontal axis is in the same direction as the edge of the image, so the angle between the connecting line and the edge of the image can also be directly obtained, that is, λ'.

[0085] Step 102, convert the third point cloud data into a second two-dimensional image located on the XZ coordinate plane, take the Y-axis data of each point as the corresponding gray value of each point on the second two-dimensional image, and identify the angle between the first reference body bottom edge, the second reference body bottom edge, the third reference body bottom edge, or the ground and the image edge as the Y-axis deflection angle of the third point cloud data.

[0086] Specifically, the point cloud of c1 through Pose side (6, 0, 3, 270, 0, 90) is obtained, attached Figure 11 is the 2d image converted from the 3d point cloud, and the angle between the arrowed connecting line and the horizontal axis in attached Figure 11 can be calculated through various existing conventional 2d image algorithms. The arrowed line can be the connecting line between the first reference body bottom edge and the second and third reference body bottom edges, or it can be the ground identification line in the figure, so as to obtain the rotation angle between w' and the y-axis of w, which is β'.

[0087] Step 103: Convert the third point cloud data into a third two-dimensional image located on the YZ coordinate plane, use the X-axis data of each point as the gray value corresponding to each point on the third two-dimensional image, and identify and obtain the angle between the bottom edge of the first reference body, the bottom edge of the second reference body, the bottom edge of the third reference body, or the ground and the edge of the image on the third two-dimensional image as the X-axis deflection angle of the third point cloud data.

[0088] Specifically, obtaining c1 via Pose back Point cloud in pose (6, 0, 10, 270, 0, 180), with appendix Figure 12 This is a 2D image converted from the 3D point cloud. The attachments can be calculated using 2D image algorithms. Figure 12 The angle between the arrowed horizontal line and the horizontal axis can be either the line connecting the bottom edge of the reference body or the ground marking line in the figure, thus obtaining the rotation angle α' between w' and the x-axis of w.

[0089] Step 104: The first angle offset matrix is ​​formed by the axis deflection angles of the obtained third point cloud data, and the first angle offset matrix is ​​combined with the first transformation matrix to form the third transformation matrix.

[0090] The above calculations show that the xyz axis angular deviations between the first coarse world coordinate system w′ and the world coordinate system w or the first precise world coordinate system are α', β', and λ', respectively. This means that the first point cloud data c1, after undergoing the third transformation matrix... The transformation yields a coordinate system w' that is in the same direction as w with high precision. Additionally, based on the position of the device's zero point in the w coordinate system, i.e., ... Figure 7 As shown in the figure, the translation distances from the zero point of the device to the origin w' are x', y'+3, and z'+6, from which the attitude of c1 transformed to w can be obtained.

[0091] Specifically, according to the above steps, in this embodiment, the first point cloud data C1 attitude is converted into the first precise world coordinate system w”, which may include the following transformation process.

[0092] Posture description:

[0093] Pose over (TransX,TransY,TransZ,RotX,RotY,RotZ)

[0094] Pose over : Indicates the pose. The subscript is a description or note of the 2D image after the pose transformation. Over indicates a top-down pose, side indicates a side-down pose, and back indicates a back-down pose.

[0095] TransX: Translation along the x-axis, positive values ​​for point cloud moving in the positive x-direction and negative values ​​for moving in the negative x-direction;

[0096] TransY: translation along y axis, positive value for y positive direction, negative value for negative direction;

[0097] TransZ: translation along z axis, positive value for z positive direction, negative value for negative direction;

[0098] RotX: rotation around x axis, positive value for clockwise rotation when x axis direction is away from the observer, unit °;

[0099] RotY: rotation around y axis, positive value for clockwise rotation when y axis direction is away from the observer, unit °;

[0100] RotZ: rotation around z axis, positive value for clockwise rotation when z axis direction is away from the observer, unit °;

[0101] The pose conversion calculation formula is:

[0102] Point Rotating around Pose(x t ,y t ,z t ,α,β,γ) pose transformation formula is as follows:

[0103]

[0104] From the above formula

[0105]

[0106]

[0107]

[0108] Each point in c1 n = 0, 1, 2, 3Λn rotates around Pose(x t ,y t ,z t ,α,β,γ) pose through the above formula conversion, the point cloud coordinate system can be changed. After c1 changes w' is obtained, w' and w have the same direction of each axis, but there is still a small angle caused by installation error.

[0109] Method for measuring the angle and position deviation between c1 and w: as described above, according to the position of the first holder, the rough conversion relationship between the first holder coordinate system c1 and the world coordinate system w Pose front (6,2,0,0,0,90), c1 through After the change, w' is obtained, w' is basically in the same direction as w on the xyz axis, but there is still a small included angle. In order to correct the angle of w' to obtain w'', the c1 is changed to Obtain a coordinate system w'' orthogonal to w with high precision.

[0110] In this embodiment, as shown in the accompanying Figure 13 The step S3 specifically further includes the following contents of the installation error angle correction of the fourth point cloud data obtained by the second laser radar on the second holder:

[0111] Step 201, convert the fourth point cloud data into a fourth two-dimensional image located on the XY coordinate plane, take the Z-axis data of each point as the corresponding gray value of each point on the fourth two-dimensional image, identify the connecting line between the intersection side of the second and third reference bodies and the axis side of the first reference body in the close area, and calculate the included angle between the connecting line and the image edge as the Z-axis deflection angle of the fourth point cloud data;

[0112] Step 202, convert the fourth point cloud data into a fifth two-dimensional image located on the XZ coordinate plane, take the Y-axis data of each point as the corresponding gray value of each point on the fifth two-dimensional image, identify the included angle between the second reference body bottom edge, the third reference body bottom edge, the fourth reference body bottom edge, or the ground and the image edge on the fifth two-dimensional image as the Y-axis deflection angle of the fourth point cloud data;

[0113] Step 203, convert the fourth point cloud data into a sixth two-dimensional image located on the YZ coordinate plane, take the X-axis data of each point as the corresponding gray value of each point on the sixth two-dimensional image, identify the included angle between the second reference body bottom edge, the third reference body bottom edge, the fourth reference body bottom edge, or the ground and the image edge on the sixth two-dimensional image as the X-axis deflection angle of the fourth point cloud data;

[0114] Step 204, combine the second angle offset matrix composed of the obtained axis deflection angles of the fourth point cloud data with the second transformation matrix to form a fourth transformation matrix.

[0115] The above steps S201 to S204 are to process and coordinate transform the point cloud data obtained by the second laser radar on the second holder, and the contents are basically the same as the processing and coordinate transformation of the point cloud data obtained by the first laser radar on the first holder in the aforementioned steps S101 to S104. Therefore, it will not be repeated and described in detail, and the specific step contents and effects can be referred to the aforementioned steps S101 to S104.

[0116] Step S4, the third point cloud data is converted into the fifth point cloud data in the first accurate world coordinate system through a third transformation matrix; the fourth point cloud data is converted into the sixth point cloud data in the second accurate world coordinate system through a fourth transformation matrix; and an offset matrix is obtained according to the positions of the same reference body in the fifth point cloud data and the sixth point cloud data.

[0117] As shown in the accompanying drawings, region ① in the figure is the ground of c1, region ② is the ground of c2, and region ③ is the intersection of the two. The four black rectangular reference bodies in the figure pass through region 1 and region 2, perfectly connecting the x-axes of the two regions, ensuring the consistency of the x-axis directions of w and w2, and also ensuring the consistency of the y-axis directions because the ground is horizontal. The xy-axes are unified as above, and the z-axis is naturally unified. Then, the 3D point cloud is converted into a 2D image again, and the coordinates of the interaction point o' in w can be measured as x″2 through a two-dimensional image processing algorithm or other conventional means. Figure 7

[0118] In another embodiment, step S4 can further include obtaining the position offset of the second reference body or the third reference body in the fifth point cloud data and the sixth point cloud data, and obtaining the offset matrix according to the offset of each axis, specifically as follows:

[0119] Step S41, the sixth point cloud data is transformed using a fifth transformation matrix to obtain seventh point cloud data, and the third reference body data in the seventh point cloud data is the same as the first reference body data in the fifth point cloud data.

[0120] Step S42, the offset distance of the first reference body and the second reference body in the X-axis is obtained, and the fifth transformation matrix is increased by the X-axis offset distance to form the offset matrix.

[0121] In the embodiment of step S4, the coordinate system is corrected by the position of the reference body. The sixth point cloud data in the second accurate coordinate system is translated into point cloud data in the first accurate coordinate system with the first reference body and the third reference body as the reference, and then the offset of the seventh point cloud data is supplemented by the offset distance x″2 of the first reference body and the second reference body in the X-axis to obtain two point cloud data in the same coordinate system.

[0122] Step S5, the point cloud data of the measured object obtained by the first laser radar is transformed through the third transformation matrix, and the point cloud data of the measured object obtained by the second laser radar is transformed through the fourth transformation matrix and the offset matrix, and then the two are superimposed to obtain the complete point cloud data of the measured object.

[0123] ​Finally, the first point cloud data and the second point cloud data can be converted to the same world coordinate system w, and then splicing and calibration can be performed. After the above various attitude transformations, 3D to 2D image, through complex coordinate conversion and formula calculation, the measured angle and displacement compensation are obtained and

[0124] The first point cloud data c1 is converted to the attitude of the world coordinate system w:

[0125]

[0126] The second point cloud data c2 is converted to the attitude of the world coordinate system w:

[0127]

[0128] The first point cloud data c1 and the second point cloud data c2 are converted by and respectively, and the splicing and calibration of the dual gimbal point cloud are completed. In this embodiment, the point cloud is converted to the w coordinate system corresponding to the actual physical.

[0129] Among the above embodiments disclosed in the measurement area provided with the first gimbal and the second gimbal, the third transformation matrix can be And the combination of the fourth transformation matrix and the offset matrix is When scanning and detecting the vehicle compartment of the vehicle to be loaded in the measurement area, the point cloud data c1 obtained by the first laser radar on the first gimbal is converted to w after The effect is shown in the accompanying Figure 14 The point cloud data c2 obtained by the second laser radar on the second gimbal is converted to w after The effect is shown in the accompanying Figure 15 After the original point cloud c1 and c2 are converted to the same world coordinate system w by and , then superimposed, the point cloud vehicle shown in the accompanying Figure 16 and 17 is obtained, which has the same size as the actual vehicle.

[0130] The front and rear partial point cloud data of a large-span object obtained by at least two laser radars respectively in a measurement area are spliced, a reference body arranged in the measurement area is used for coordinate system conversion correction in the early stage, the attitude of a virtual camera for 3D conversion to 2D is corrected, and the angle deviation between the gimbal coordinate system and the world coordinate system is obtained. Through a common object reference plane, the mutual relationship between the two gimbal coordinate systems is obtained, and finally the coordinate systems of the two gimbals are converted to the same device coordinate system, the point clouds collected by the two gimbals are spliced to form point cloud data of an ultra-large size, and a small part of the point cloud data in the point cloud data obtained by the two laser radars is overlapped, so that the point clouds collected by the two gimbals can be spliced to form the point cloud data of an ultra-large size. The method relies on two independent gimbals to scan two scenes separately, and only a small part of the overlapped point cloud in the scene is used to determine the relationship between the two gimbal coordinate systems, and realize point cloud splicing. Of course, the method is not limited to the splicing of point cloud data of two laser radars, but can also be applied to the splicing combination of point cloud data obtained by three or more independent gimbals.

[0131] In another embodiment, a three-dimensional detection system for acquiring three-dimensional data of a large-span object is also disclosed, comprising at least two laser radars installed in a measurement area, and a controller in communication with the laser radars, the laser radars being used to acquire partial point cloud data of front and back parts of the large-span object respectively and send to the controller, the controller being configured to: acquire first point cloud data containing a reference body in a first gimbal coordinate system through the first laser radar, and acquire second point cloud data containing the reference body in a second gimbal coordinate system through the second laser radar. A first transformation matrix is acquired according to the installation position of the first laser radar, and the first point cloud data is converted into third point cloud data in a first rough world coordinate system through the first transformation matrix; a second transformation matrix is acquired according to the installation position of the second laser radar, and the second point cloud data is converted into fourth point cloud data in a second rough world coordinate system through the second transformation matrix. The third point cloud data is converted into two-dimensional images located in three coordinate planes respectively, a first angle offset matrix is acquired according to the offset angle of the reference body in the two-dimensional images, and the first angle offset matrix is combined with the first transformation matrix to form a third transformation matrix; the fourth point cloud data is converted into two-dimensional images located in three coordinate planes respectively, a second angle offset matrix is acquired according to the offset angle of the reference body in the two-dimensional images, and the second angle offset matrix is combined with the second transformation matrix to form a fourth transformation matrix. The third point cloud data is converted into fifth point cloud data in a first accurate world coordinate system through the third transformation matrix; the fourth point cloud data is converted into sixth point cloud data in a second accurate world coordinate system through the fourth transformation matrix; and an offset matrix is acquired according to the positions of the same reference body in the fifth point cloud data and the sixth point cloud data. The point cloud data of the measured object acquired by the first laser radar is transformed through the third transformation matrix, and the point cloud data of the measured object acquired by the second laser radar is transformed through the fourth transformation matrix and the offset matrix, and then the transformed point cloud data is superimposed to obtain complete point cloud data of the measured object.

[0132] In the embodiment, the controller is configured to: acquire first point cloud data containing first, second and third reference bodies in a gimbal coordinate system through the first laser radar; the first to fourth reference bodies are placed in sequence along the central axis of the area in the measurement area, wherein the second reference body and the third reference body are located in the measurement intersection area of the two laser radars, and the side of the second reference body and the side of the third reference body opposite to the side of the second reference body are located on the central axis of the area, the back side of the second reference body and the front side of the third reference body are located on a vertical plane perpendicular to the central axis of the area, and the first and fourth reference bodies are located in the first laser radar measurement area and the second laser radar measurement area respectively, and the same side of the first and fourth reference bodies is located on the central axis of the area; acquire second point cloud data containing the second, third and fourth reference bodies in the gimbal coordinate system through the second laser radar.

[0133] In the embodiment, the controller is further configured to: convert the third point cloud data into a first two-dimensional image located on an XY coordinate plane, take the Z-axis data of each point as the corresponding gray value of each point on the first two-dimensional image, identify a line connecting the intersection side of the second and third reference bodies and the axis line side of the close area of the first reference body in the first two-dimensional image, and calculate the included angle between the line and the image edge as the Z-axis deflection angle of the third point cloud data. Convert the third point cloud data into a second two-dimensional image located on an XZ coordinate plane, take the Y-axis data of each point as the corresponding gray value of each point on the second two-dimensional image, identify the included angle between the bottom edge of the first reference body, the bottom edge of the second reference body, the bottom edge of the third reference body, or the ground and the image edge on the second two-dimensional image as the Y-axis deflection angle of the third point cloud data. Convert the third point cloud data into a third two-dimensional image located on a YZ coordinate plane, take the X-axis data of each point as the corresponding gray value of each point on the third two-dimensional image, identify the included angle between the bottom edge of the first reference body, the bottom edge of the second reference body, the bottom edge of the third reference body, or the ground and the image edge on the third two-dimensional image as the X-axis deflection angle of the third point cloud data. The obtained axis deflection angles of the third point cloud data form a first angle offset matrix, and the first angle offset matrix is combined with the first transformation matrix to form a third transformation matrix.

[0134] In the embodiment, the controller is further configured to: convert the fourth point cloud data into a fourth two-dimensional image located on an XY coordinate plane, take the Z-axis data of each point as the corresponding gray value of each point on the fourth two-dimensional image, identify a line connecting the intersection side of the second and third reference bodies and the axis line side of the close area of the first reference body in the fourth two-dimensional image, and calculate the included angle between the line and the image edge as the Z-axis deflection angle of the fourth point cloud data. Convert the fourth point cloud data into a fifth two-dimensional image located on an XZ coordinate plane, take the Y-axis data of each point as the corresponding gray value of each point on the fifth two-dimensional image, identify the included angle between the bottom edge of the second reference body, the bottom edge of the third reference body, the bottom edge of the fourth reference body, or the ground and the image edge on the fifth two-dimensional image as the Y-axis deflection angle of the fourth point cloud data. Convert the fourth point cloud data into a sixth two-dimensional image located on a YZ coordinate plane, take the X-axis data of each point as the corresponding gray value of each point on the sixth two-dimensional image, identify the included angle between the bottom edge of the second reference body, the bottom edge of the third reference body, the bottom edge of the fourth reference body, or the ground and the image edge on the sixth two-dimensional image as the X-axis deflection angle of the fourth point cloud data. The obtained axis deflection angles of the fourth point cloud data form a second angle offset matrix, and the second angle offset matrix is combined with the second transformation matrix to form a fourth transformation matrix.

[0135] It should be noted that the various embodiments described in the specification are progressive in nature, and each embodiment focuses on the differences from other embodiments. The same or similar parts between embodiments can be mutually referred to. For the controller of the three-dimensional detection system disclosed in the embodiments, since it corresponds to the large-span object three-dimensional point cloud data combination method disclosed in the embodiments, the description is relatively simple, and the relevant parts can be referred to the method part.

[0136] The application further discloses a large-span object three-dimensional point cloud data combination device for splicing front and rear partial point cloud data of a large-span object obtained by two laser radars installed in a measurement area, respectively. The device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the large-span object three-dimensional point cloud data combination method of any one of the preceding embodiments are implemented.

[0137] For example, the computer program can be divided into one or more modules / units, which are stored in the memory and executed by the processor to complete the application. The one or more modules / units can be a series of computer program instruction segments capable of completing a specific function, which are used to describe the execution process of the computer program in the server.

[0138] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor, etc. The processor is the control center of the server device, and connects various parts of the server device through various interfaces and lines.

[0139] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the server device by running or executing the computer program and / or modules stored in the memory, and calling data stored in the memory. The memory can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function, and the like. In addition, the memory can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0140] If the cargo stacking control method of the mechanical arm is implemented in the form of a software function unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by instructing related hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the processor executes the computer program, the steps of the above-mentioned various method embodiments can be realized. The computer program includes computer program code, which can be in the form of source code, object code, an executable file, or some intermediate form, etc. The computer-readable medium can include any entity or device capable of carrying the computer program code, a recording medium, a U disk, a mobile hard disk, a magnetic disk, an optical disk, a computer memory, a read-only memory (ROM), a random access memory (RAM), an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content included in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer-readable medium does not include electrical carrier signals and telecommunication signals.

[0141] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

[0142] In summary, the above is only the preferred embodiment of the present application, any equivalent changes and modifications made in accordance with the scope of the present application patent application should be covered by the present application patent.

Claims

1. A method for combining three-dimensional point cloud data of a large-span object, used to stitch together local point cloud data of the front and rear parts of a large-span object acquired by at least two lidar sensors installed within a measurement area, characterized in that... Includes the following steps: S1, acquiring first point cloud data containing a reference body in the first gimbal coordinate system using a first lidar, and acquiring second point cloud data containing a reference body in the second gimbal coordinate system using a second lidar; step S1 specifically includes: S11, the first point cloud data containing the first, second, and third reference bodies in the gimbal coordinate system is acquired by the first lidar; the first to fourth reference bodies are placed sequentially along the central axis of the region within the measurement area, wherein the second and third reference bodies are located within the measurement intersection area of ​​the two lidars, and one side of the second reference body and the other side of the third reference body relative to the second reference body are respectively located on the central axis of the region, the rear side of the second reference body and the front side of the third reference body are respectively located on a vertical plane perpendicular to the central axis of the region, and the first and fourth reference bodies are respectively located within the measurement areas of the first lidar and the second lidar, which do not intersect, and the same side of the first and fourth reference bodies is located on the central axis of the region; S12, acquire the second point cloud data containing the second, third and fourth reference bodies in the gimbal coordinate system through the second lidar; S2, obtain the first transformation matrix according to the installation position of the first lidar, and convert the first point cloud data into third point cloud data in the first coarse world coordinate system through the first transformation matrix; The second transformation matrix is ​​obtained based on the installation location of the second lidar, and the second point cloud data is converted into fourth point cloud data in the second coarse world coordinate system using the second transformation matrix. S3, convert the third point cloud data into two-dimensional images located in three coordinate planes respectively, obtain the first angle offset matrix according to the reference body offset angle in the two-dimensional image, and combine the first angle offset matrix with the first transformation matrix to form the third transformation matrix; The fourth point cloud data is converted into two-dimensional images located in three coordinate planes. The second angle offset matrix is ​​obtained based on the offset angle of the reference body in the two-dimensional image. The second angle offset matrix is ​​combined with the second transformation matrix to form the fourth transformation matrix. The third point cloud data is converted into a first two-dimensional image located on the XY coordinate plane. The Z-axis data of each point is used as the gray value corresponding to each point on the first two-dimensional image. The line connecting the first reference body in the first two-dimensional image to the side of the central axis near the region and the intersection of the second and third reference bodies is identified and obtained. The angle between the connecting line and the edge of the image is calculated and obtained as the Z-axis deflection angle of the third point cloud data. The third point cloud data is converted into a second two-dimensional image located on the XZ coordinate plane. The Y-axis data of each point is used as the gray value corresponding to each point on the second two-dimensional image. The angle between the bottom edge of the first reference body, the bottom edge of the second reference body, the bottom edge of the third reference body, or the ground and the edge of the image on the second two-dimensional image is identified and obtained as the Y-axis deflection angle of the third point cloud data. The third point cloud data is converted into a third two-dimensional image located on the YZ coordinate plane. The X-axis data of each point is used as the gray value corresponding to each point on the third two-dimensional image. The angle between the bottom edge of the first reference body, the bottom edge of the second reference body, the bottom edge of the third reference body, or the ground and the edge of the image on the third two-dimensional image is identified and obtained as the X-axis deflection angle of the third point cloud data. The first angle offset matrix is ​​formed by the deflection angles of each axis of the obtained third point cloud data. The first angle offset matrix is ​​combined with the first transformation matrix to form the third transformation matrix. S4, transform the third point cloud data into the fifth point cloud data in the first precise world coordinate system using the third transformation matrix; transform the fourth point cloud data into the sixth point cloud data in the second precise world coordinate system using the fourth transformation matrix; obtain the offset matrix based on the position of the same reference body in the fifth and sixth point cloud data; S5, the point cloud data of the object under test acquired by the first lidar is transformed by the third transformation matrix, and then superimposed with the point cloud data of the object under test acquired by the second lidar after being transformed by the fourth transformation matrix and the offset matrix to obtain the complete point cloud data of the object under test.

2. The method for combining three-dimensional point cloud data of a large-span object according to claim 1, step S3 further includes: The fourth point cloud data is converted into a fourth two-dimensional image located on the XY coordinate plane. The Z-axis data of each point is used as the gray value corresponding to each point on the fourth two-dimensional image. The line connecting the intersection of the second and third reference bodies in the fourth two-dimensional image and the side of the first reference body near the central axis of the region is identified and obtained. The angle between the line and the edge of the image is calculated and obtained as the Z-axis deflection angle of the fourth point cloud data. The fourth point cloud data is converted into a fifth two-dimensional image located on the XZ coordinate plane. The Y-axis data of each point is used as the gray value corresponding to each point on the fifth two-dimensional image. The bottom edge of the second reference body, the bottom edge of the third reference body, the bottom edge of the fourth reference body, or the angle between the ground and the edge of the image on the fifth two-dimensional image are identified and obtained as the Y-axis deflection angle of the fourth point cloud data. The fourth point cloud data is converted into a sixth two-dimensional image located on the YZ coordinate plane. The X-axis data of each point is used as the gray value corresponding to each point on the sixth two-dimensional image. The bottom edge of the second reference body, the bottom edge of the third reference body, the bottom edge of the fourth reference body, or the angle between the ground and the edge of the image on the sixth two-dimensional image are identified and obtained as the X-axis deflection angle of the fourth point cloud data. The deflection angles of each axis of the obtained fourth point cloud data are used to form the second angle offset matrix. The second angle offset matrix is ​​combined with the second transformation matrix to form the fourth transformation matrix.

3. The method for combining three-dimensional point cloud data of a large-span object according to claim 2, wherein step S4 further includes: Obtain the position offset of the second or third reference body in the fifth and sixth point cloud data, and obtain the offset matrix based on the offset of each axis.

4. A three-dimensional detection system for acquiring three-dimensional data of large-span objects, characterized in that, The system includes at least two lidar units installed within the measurement area and a controller communicating with the lidar units. The lidar units acquire local point cloud data of the front and rear portions of a large-span object and send it to the controller. The controller is configured to: The system acquires first point cloud data containing reference bodies in a first gimbal coordinate system using a first lidar, and second point cloud data containing reference bodies in a second gimbal coordinate system using a second lidar. The system also acquires first point cloud data containing first, second, and third reference bodies in a gimbal coordinate system using the first lidar. The first to fourth reference bodies are placed sequentially along the central axis of the measurement area. The second and third reference bodies are located within the intersection area of ​​the two lidar measurements, with one side of the second reference body and the other side of the third reference body relative to the second reference body located on the central axis of the area. The rear side of the second reference body and the front side of the third reference body are located on vertical planes perpendicular to the central axis of the area. The first and fourth reference bodies are located within the non-intersecting measurement areas of the first and second lidars, with the same side of the first and fourth reference bodies located on the central axis of the area. The system also acquires second point cloud data containing second, third, and fourth reference bodies in a gimbal coordinate system using the second lidar. The first transformation matrix is ​​obtained based on the installation location of the first lidar, and the first point cloud data is converted into third point cloud data in the first coarse world coordinate system using the first transformation matrix; the second transformation matrix is ​​obtained based on the installation location of the second lidar, and the second point cloud data is converted into fourth point cloud data in the second coarse world coordinate system using the second transformation matrix. The third point cloud data is converted into two-dimensional images located in three coordinate planes. A first angular offset matrix is ​​obtained based on the offset angle of the reference body in the two-dimensional image. This first angular offset matrix is ​​combined with a first transformation matrix to form a third transformation matrix. The fourth point cloud data is also converted into two-dimensional images located in three coordinate planes. A second angular offset matrix is ​​obtained based on the offset angle of the reference body in the two-dimensional image. This second angular offset matrix is ​​combined with a second transformation matrix to form a fourth transformation matrix. The third point cloud data is then converted into a first two-dimensional image located in the XY coordinate plane. The Z-axis data of each point is used as the grayscale value corresponding to each point in the first two-dimensional image. The line connecting the first reference body near the central axis of the region in the first two-dimensional image to the intersection of the second and third reference bodies is identified. The angle between this line and the image edge is calculated as the Z-axis deflection angle of the third point cloud data. The third point cloud data is then... The data is converted into a second two-dimensional image located on the XZ coordinate plane. The Y-axis data of each point is used as the gray value corresponding to each point on the second two-dimensional image. The angle between the bottom edge of the first reference body, the bottom edge of the second reference body, the bottom edge of the third reference body, or the ground and the edge of the image on the second two-dimensional image is identified and obtained as the Y-axis deflection angle of the third point cloud data. The third point cloud data is converted into a third two-dimensional image located on the YZ coordinate plane. The X-axis data of each point is used as the gray value corresponding to each point on the third two-dimensional image. The angle between the bottom edge of the first reference body, the bottom edge of the second reference body, the bottom edge of the third reference body, or the ground and the edge of the image on the third two-dimensional image is identified and obtained as the X-axis deflection angle of the third point cloud data. The obtained axis deflection angles of the third point cloud data are used to form a first angle offset matrix. The first angle offset matrix is ​​combined with the first transformation matrix to form a third transformation matrix. The third point cloud data is transformed into the fifth point cloud data in the first precise world coordinate system using the third transformation matrix; The fourth point cloud data is transformed into the sixth point cloud data in the second precise world coordinate system using the fourth transformation matrix; the offset matrix is ​​obtained based on the position of the same reference body in the fifth and sixth point cloud data. The point cloud data of the object under test acquired by the first lidar is transformed by the third transformation matrix and then superimposed with the point cloud data of the object under test acquired by the second lidar after being transformed by the fourth transformation matrix and the offset matrix to obtain the complete point cloud data of the object under test.

5. The three-dimensional detection system according to claim 4, characterized in that, The controller is configured to: The fourth point cloud data is converted into a fourth two-dimensional image located on the XY coordinate plane. The Z-axis data of each point is used as the gray value corresponding to each point on the fourth two-dimensional image. The line connecting the intersection of the second and third reference bodies in the fourth two-dimensional image and the side of the first reference body near the central axis of the region is identified and obtained. The angle between the line and the edge of the image is calculated and obtained as the Z-axis deflection angle of the fourth point cloud data. The fourth point cloud data is converted into a fifth two-dimensional image located on the XZ coordinate plane. The Y-axis data of each point is used as the gray value corresponding to each point on the fifth two-dimensional image. The bottom edge of the second reference body, the bottom edge of the third reference body, the bottom edge of the fourth reference body, or the angle between the ground and the edge of the image on the fifth two-dimensional image are identified and obtained as the Y-axis deflection angle of the fourth point cloud data. The fourth point cloud data is converted into a sixth two-dimensional image located on the YZ coordinate plane. The X-axis data of each point is used as the gray value corresponding to each point on the sixth two-dimensional image. The bottom edge of the second reference body, the bottom edge of the third reference body, the bottom edge of the fourth reference body, or the angle between the ground and the edge of the image on the sixth two-dimensional image are identified and obtained as the X-axis deflection angle of the fourth point cloud data. The deflection angles of each axis of the obtained fourth point cloud data are used to form the second angle offset matrix. The second angle offset matrix is ​​combined with the second transformation matrix to form the fourth transformation matrix.

6. A device for combining three-dimensional point cloud data of a large-span object, used to stitch together local point cloud data of the front and rear parts of a large-span object acquired by two lidars installed in a measurement area, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, it implements the steps of the method as described in any one of claims 1-3.

Images