CN107703499A - A kind of point cloud error calibration method based on self-control ground laser radar alignment error - Google Patents

Abstract

The invention discloses a point cloud error correction method based on the self-made ground-based laser radar alignment error. The method is mainly aimed at the scanning mode of the self-made laser radar system with a 45° rotating mirror and a pan-tilt rotation, by calculating the alignment between the incident light and the pitch axis. The influence of the alignment error on the angle measurement error of the self-made system, the point cloud error correction model of the self-made lidar system is established, and the point cloud coordinates of the target scanned by the high-precision 3D scanner are used as the real value to solve the alignment error, so as to realize the alignment error of the self-made lidar system The error correction of the quasi-error to the target point cloud. The method mainly includes the following three steps: 1) establishing and simplifying the angle measurement error model of the self-made LiDAR 3D imaging system caused by alignment errors; 2) obtaining the point cloud error model of the self-made LiDAR 3D imaging system; 3) solving the self-made The alignment error in the point cloud error model of the LiDAR 3D imaging system, and correct the target point cloud of the self-made LiDAR system according to the model.

Description

A point cloud error correction method based on self-made ground-based lidar alignment error

technical field

The invention relates to the technical field of laser radar measurement, in particular to a point cloud error correction method based on self-made ground-based laser radar alignment error

Background technique

Lidar measurement technology is a new active remote sensing technology that has developed rapidly in recent years. It generally uses non-contact measurement technology and has a wide range of applications in remote sensing, military detection, oceanographic mapping, and atmospheric exploration. Based on the traditional single-point measurement technology, the three-dimensional scanning measurement technology in LiDAR can quickly obtain high-resolution point cloud data on the surface of the target object through high-speed laser scanning. This technology has the advantages of simple data processing, rapidity, It has the advantages of initiative, strong anti-interference ability, high measurement accuracy and large range. However, the scanning accuracy of lidar is largely affected by the accuracy of its own instrument. In actual use, the accuracy of the instrument itself does not fully meet its nominal accuracy, or the loss caused by external force collisions during use, changes in external conditions, and long-term use And other unknown factors cause unstable performance of the instrument, and systematic errors may appear in the scanning results. Therefore, effectively eliminating the error of the lidar instrument is the key to improving the accuracy of the scanning point cloud.

According to the existing research results, lidar is mainly divided into two categories: airborne lidar and ground lidar. The point cloud error correction methods of lidar 3D imaging systems are also different. The first type is based on airborne lidar system. The calibration method is mainly based on flight self-calibration. The basic idea is to use laser to scan known target points or relative target points, and estimate the parameters of the fixed offset generated in the process, including overlapping strip calibration technology, least squares Adjustment geometric calibration technology, etc. Among them, the calibration technology of overlapping flight belts has higher requirements on the technology of flight belt splicing; the assumptions and preconditions in the solution of the adjustment method are generally difficult to establish, which may cause problems such as a decrease in the accuracy of parameter estimation. There is no comparison between the calibration method of the above-mentioned airborne laser radar system and the present invention, and there is no reference.

The second type of calibration methods based on ground-based lidar systems are mostly done indoors or in the baseline field. Some calibration experiments involving point accuracy mainly use the method of public point conversion, and most of them use the matching targets of the system for testing, including space Length detection method, self-test calibration method, etc. The space length detection method has high requirements on the detection field, requires the site to have high precision, and relies more on the accuracy of the scanning target layout; the self-inspection calibration method generally uses total stations, theodolites and other instruments to establish calibration models for the laser radar system, and the model It generally includes parameters such as the rotation angle of the coordinate system, the translation amount, and the internal error of the instrument. These parameters are obtained as unknown quantities through a unified solution. This method can continuously improve the error model by adding parameters, improve the correction accuracy of the system error, and check The scope of the lidar system is relatively wide, and there is no very high requirement for the layout accuracy of the target object. The invention belongs to a system self-checking and calibrating method.

According to the patent review provided by the existing patent office, the self-checking and calibration methods of the terrestrial lidar system are mainly divided into the following two categories: the first category uses public marker points to obtain the coordinate rotation and translation between the system to be calibrated and the high-precision scanner directly take the difference between the measured value of the system to be calibrated and the standard measured value as the measurement error of the system space coordinates. This type of method obtains the relative error of spatial coordinate measurement for all spatial three-dimensional coordinate measurement systems. Its disadvantage is that it only evaluates and corrects errors at the three-dimensional point cloud level, does not build a system error model, and requires a large amount of sample data to achieve accurate correction; the above patent Including "a calibration method for an electronic theodolite space coordinate measuring system" disclosed in Chinese patent 200810147441.9. The second type of method uses statistical methods to analyze the probability density distribution of the system ranging error and angle measuring error, and obtains the error correction samples in the three-dimensional coordinate system. This type of method obtains a system error correction model for all three-dimensional coordinate measurement systems. Its disadvantage is that the parameters of the system error model have no physical meaning, and it does not conduct theoretical analysis on the scanning mode of the system itself from the perspective of error sources. Therefore, it depends on large sample data to achieve Accurate correction; the above-mentioned patents include "a correction method for long-distance scanning laser radar measurement error" disclosed in Chinese patent 201710014687.8.

The main difference between the present invention and the existing system self-checking and calibration method patents is that it proposes a theoretical analysis of the scanning mode of the system itself from the perspective of error sources, and uses a small number of target points to realize self-made ground-based lidar alignment error point cloud error correction The method is applicable to all 3D scanning systems that use a 45° rotating mirror and a pan-tilt rotating scanning method. The invention has broad application prospects in the technical field of laser radar measurement.

Contents of the invention

The invention discloses a point cloud error correction method based on the self-made ground-based laser radar alignment error, which is characterized in that the method is aimed at the scanning mode of the 45° rotating mirror and the pan-tilt rotation in the self-made laser radar three-dimensional imaging system, wherein the self-made The laser radar three-dimensional imaging system includes an optical system, a scanning mechanism (motor, 45 ° rotating mirror), and a cloud platform; the incident light is emitted to the center (O point) of the 45 ° rotating mirror through the optical system, and the scanning mechanism is vertical Rotation, the horizontal rotation of the platform are emitted from the self-made laser radar three-dimensional imaging system; the coordinate system (O-XYZ) of the self-made laser radar three-dimensional imaging system includes a pitch axis (X axis), an initial outgoing light direction (Y axis) and azimuth axis (Z axis); ideally, the incident light coincides with the pitch axis (X axis), but in practice there is a gap between the incident light and the pitch axis (X axis) Angle β, there is an angle between the opposite vector projected by the incident light on the YOZ plane and the azimuth axis (Z axis) The alignment error is defined as: ① the angle β between the actual incident light and the pitch axis (X axis); ② the opposite vector of the actual incident light projected on the YOZ plane and the azimuth Angle between axes (Z-axis) Counting from the positive semi-axis of the Z axis, the clockwise direction is positive, and the range is 0° to 360°; the one-dimensional distance L of a point P on the target in the coordinate system (O-XYZ) is defined as length, azimuth defined as The angle between the projection on the XOY plane and the X-axis is calculated from the positive semi-axis of the X-axis, and the counterclockwise direction is positive, the range is 0° to 360°, and the pitch angle defined as 90° with and the angle difference between the Z axis, the one-dimensional distance L, the azimuth angle and the pitch angle The measurement errors are ranging error ΔL, azimuth angle error and pitch error The alignment error is analyzed by theoretical For the self-made laser radar three-dimensional imaging system angle measurement error (azimuth angle error pitch angle error ), the angle measurement error model of the self-made lidar three-dimensional imaging system is established; the point cloud of the target is defined as the Cartesian coordinates (x, y, z) ^T of a point on the target, and the point cloud error of the target is defined as the The deviation (Δx, Δy, Δz) ^T between the coordinate measurement value of the target point cloud and the true value, according to the point cloud error (Δx, Δy, Δz) ^T of the target and the angle measurement error (azimuth angle error) of the target pitch angle error ), the point cloud error model of the self-made lidar three-dimensional imaging system is established as follows:

Obtain the homemade lidar 3D imaging system for the alignment error The corrected target point cloud coordinates are:

(x+Δx,y+Δy,z+Δz) ^T

Thereby realizing the alignment error of the self-made lidar 3D imaging system Point cloud error correction;

The method mainly includes the following seven steps:

1) Establish the coordinate system (O-XYZ) of the self-made laser radar three-dimensional imaging system; the measured value of any point P on the target in the coordinate system (O-XYZ) is the azimuth angle Pitch angle The true value of any point P on the target in the coordinate system (O-XYZ) is the azimuth Pitch angle said with The difference is the azimuth error said with The difference is the pitch angle error

2) Establish the alignment error by the The angle measurement error model of the self-made laser radar three-dimensional imaging system caused; This model describes the angle measurement error (azimuth angle error) respectively pitch angle error ) and the alignment error and the azimuth of the target Pitch angle The mathematical relationship between them is as follows:

3) Based on the alignment error for the The value of the trigonometric function is approximately transformed, and the angle measurement error model of the self-made lidar 3D imaging system is simplified, as follows:

4) According to the principle of error transmission, establish the point cloud error (Δx, Δy, Δz) ^T of the self-made lidar three-dimensional imaging system and the distance measurement error ΔL, the angle measurement error (azimuth error pitch angle error ) between the mapping relationship, as follows:

Wherein, L is the one-dimensional distance of any point on the target in the self-made laser radar three-dimensional imaging system, and ΔL is the measurement error of the one-dimensional distance L of any point on the target;

5) The angle measurement error model of the self-made laser radar three-dimensional imaging system is transformed into a Cartesian coordinate system, and the point cloud error model of the self-made laser radar three-dimensional imaging system is obtained as follows:

6) Using the self-made laser radar three-dimensional imaging system to scan the N targets, and obtain the coordinates ( _xi , y _i , z _i ) ^T of the target in the self-made laser radar three-dimensional imaging system, (i =1,2,...,N), as the measured value; use the high-precision three-dimensional scanner to scan the target twice to obtain the coordinates of the target in the high-precision three-dimensional scanner (x′ _i , y ′ _i , z′ _i ) ^T , (i=1,2,…,N), transform it into the self-made lidar three-dimensional imaging system as the real value; the difference between the measured value and the real value is the point cloud error (Δx _i , Δy _i , Δz _i ) ^T of the self-made lidar three-dimensional imaging system, (i=1,2,...,N); according to the mapping relationship between the spherical coordinate system and the rectangular coordinate system , get the one-dimensional distance L _i and azimuth of the target from the measured value and pitch angle The ranging error ΔL _i of the target is regarded as a known constant when the one-dimensional distance L _i is within a certain range;

7) Solve the model parameters in the point cloud error model of the self-made lidar three-dimensional imaging system, that is, the alignment error Alignment error of the self-made lidar 3D imaging system The point cloud error of the _self -made _{lidar three} -dimensional imaging system is ^corrected ; distance L _i , azimuth angle Pitch angle and ranging error ΔL _i are substituted into the formula (4), 3*N nonlinear equations are obtained, and the alignment in the self-made lidar three-dimensional imaging system point cloud error model is determined by solving the nonlinear equations error The alignment error Substituting into formula (4), on the basis of the target point cloud coordinates (x, y, z) ^T of the self-made laser radar three-dimensional imaging system, the corrected target point cloud of the self-made laser radar three-dimensional imaging system is obtained Coordinates (x+Δx,y+Δy,z+Δz) ^T .

Wherein, set up described self-made laser radar three-dimensional imaging system coordinate system (O-XYZ); The pitch axis of described self-made laser radar three-dimensional imaging system is defined as the motor shaft of the 45 ° rotating mirror, and the azimuth axis is defined as the cloud platform The axis of rotation; ideally, the incident light coincides with the pitch axis; the center of the reflective surface of the 45° rotating mirror, that is, the intersection point where the ideal incident light is incident on the reflective surface is the coordinate origin O; and The pitch axes coincide, and the coordinate axis whose positive direction is the same as the direction in which the ideal incident light is incident on the 45° rotating mirror is defined as the X axis; the coordinate axis that is the same as the initial outgoing direction of the ideal incident light is defined as Y Axis; coincident with the azimuth axis, the coordinate axis vertically upward in the positive direction is defined as the Z axis.

where, established by the alignment error The angle measurement error model of the self-made lidar three-dimensional imaging system caused by it is simplified; because there is an angle β between the actual incident light and the pitch axis (X axis), the actual incident light is in the YOZ plane There is an angle between the opposite vector of the upward projection and the azimuth axis (Z axis) Counting from the positive semi-axis of the Z axis, the clockwise direction is positive, and the range is 0° to 360°;

The establishment process of the self-made laser radar three-dimensional imaging system angle measurement error model is as follows:

The initial normal unit vector of the 45° rotating mirror (under the O-XYZ coordinate system) is:

Ideally, the outgoing ray passes through a point P on the target. At this time, the normal vector of the 45° rotating mirror (under the O-XYZ coordinate system) is:

In formula (6) is the measured value of the pitch angle of a point P on the target, is the azimuth measurement value of a point P on the target, are respectively rotated counterclockwise around the X-axis by the angle, rotate the Z axis counterclockwise the The rotation matrix of the angle, the rotation matrix is as follows:

Substituting formula (5) and formula (7) into formula (6), the normal vector of the 45 ° rotating mirror under the O-XYZ coordinate system can be obtained as follows:

Initially the actual incident light vector is After the pan-tilt rotates, it becomes The vector form of the light reflection law is Find the outgoing light vector as According to the conversion relationship between spherical coordinates and rectangular coordinates, as shown in the following formula (9), the true values of the azimuth angle and the pitch angle can be obtained

According to the principle of trigonometric function approximation, the azimuth error The pitch error with the It can be approximated as follows:

The obtained angle measurement error model of the self-made lidar three-dimensional imaging system includes the azimuth error and pitch error describes the angular error (azimuth error pitch angle error ) and the alignment error and the azimuth The pitch angle the mathematical relationship between

According to the trigonometric function approximation principle sinβ=β, cosβ=1, and reasonably omit some high-order terms of β, the angle measurement error model of the self-made lidar 3D imaging system after simplification is obtained as follows:

Wherein, the angle measurement error model in the spherical coordinate system of the self-made laser radar three-dimensional imaging system is converted to the point cloud error model in the rectangular coordinate system of the self-made laser radar three-dimensional imaging system; the self-made laser radar three-dimensional imaging system can scan directly obtain the one-dimensional distance L and the azimuth of the point P on the target with the pitch angle The measured value is converted to the three-dimensional coordinates (x, y, z) ^T of a point P on the target under the Cartesian coordinate system of the self-made laser radar three-dimensional imaging system according to formula (9);

Due to the existence of the distance measurement error ΔL and the angle measurement error (azimuth angle error pitch angle error ), the coordinate measurement value of a point P on the target is P'(x+Δx, y+Δy, z+Δz), where (Δx, Δy, Δz) ^T is the point cloud error of the target, according to the error transfer Formula, the point cloud error model of the self-made lidar 3D imaging system is obtained as follows:

Wherein, the coordinates of the target in the high-precision three-dimensional scanner are converted into the self-made laser radar three-dimensional imaging system; the alignment error in the point cloud error model of the self-made laser radar three-dimensional imaging system Unknown and difficult to measure, so use the self-made lidar 3D imaging system and the point cloud coordinate data in the high-precision 3D scanner to solve the alignment error Realize point cloud error correction; Utilize the self-made laser radar three-dimensional imaging system to scan N described targets, obtain the coordinates (x _i , y _i , z _i ) ^T , (i=1,2,…,N), as measured values; substituting the measured values into formula (9), to obtain the one-dimensional distance L _i and pitch angle of all the targets and azimuth Using the high-precision three-dimensional scanner to scan the same target, obtain the coordinates (x′ _i , y′ _i , z′ _i ) ^T of a point on the target in the coordinate system of the high-precision three-dimensional scanner, (i=1,2,...,N); Since the ( _xi ,y _i , _zi ) ^T and the (x' _i ,y' _i ,z' _i ) ^T are in different coordinate systems, Therefore, the (x' _i , y' _i , z' _i ) ^T is converted to the coordinate system of the self-made laser radar three-dimensional imaging system as a real value; The rotation matrix R and translation vector T of the imaging system are expressed as follows:

In formulas (13) and (14), a, b, and c are the counterclockwise rotation angles around each axis of the coordinate system of the high-precision three-dimensional scanner, and x ₀ , y ₀ , and z ₀ are the X, Y, and The displacement in the three directions of Z, the coordinates of the (x' _i , y' _i , z' _i ) ^T in the self-made laser radar three-dimensional imaging system are as follows:

Wherein, solving the model parameters in the point cloud error model of the self-made lidar three-dimensional imaging system, that is, the alignment error The point cloud error model of the self-made lidar 3D imaging system can be written as follows:

The ranging error ΔL _i is mainly affected by the ranging circuit in the self-made laser radar three-dimensional imaging system, and is a known constant when the one-dimensional distance L _i is within a certain range; the ( _xi , y _i , z _i ) ^T , the (x′ _i ,y′ _i , z′ _i ) ^T , the one-dimensional distance L _i , the azimuth and the pitch angle Substituting formula (15) and formula (16), and solving 3*N nonlinear equations based on the least square method, the rotation matrix R and the translation vector T(a, b, c, x ₀ , y ₀ , z ₀ ) with the alignment error There are 8 unknowns in total.

Among them, using the alignment error obtained by solving Improve the point cloud error model of the self-made laser radar three-dimensional imaging system, realize the point cloud error correction of the alignment error of the self-made laser radar three-dimensional imaging system; Substituting formula (11) and formula (12), the point cloud error (Δx, Δy, Δz) ^T of the self-made lidar 3D imaging system is obtained, and the target point cloud coordinates ( On the basis of x, y, z) ^T , obtain the alignment error of the self-made laser radar three-dimensional imaging system The corrected target point cloud coordinates (x+Δx, y+Δy, z+Δz) ^T .

Wherein, the target used for point cloud error correction of alignment error of the self-made lidar 3D imaging system includes, but is not limited to, all objects that can obtain the spatial coordinates of a point on the target, such as a target ball and a plane reflective target.

Description of drawings

Figure 1 is a spatial schematic diagram of the self-made laser radar 3D imaging system scanning the target;

Figure 2 is a schematic diagram of the point cloud error correction process for the self-made ground-based lidar alignment error;

Figure 3 is a schematic diagram of the alignment error between the incident light and the pitch axis in the self-made system coordinate system;

Figure 4 is the simulation result of the angle measurement error change during the full-scale scanning process of the self-made system;

Figure 5 is the comparison result of the angle measurement error simulation before and after the simplification of the self-made system angle measurement error model;

Fig. 6 is a schematic flow diagram of parameter establishment in the self-made system point cloud error model;

Fig. 7 is a schematic diagram of the scanning scheme established by parameters in the self-made system point cloud error model;

detailed description

The specific implementation manner of the patent of the present invention will be further described in detail below in conjunction with the accompanying drawings. The flow diagram of the point cloud error correction method based on the self-made ground-based lidar alignment error is shown in Figure 2. First, the self-made lidar 3D imaging system (713) coordinate system O-XYZ is established, because the actual incident light in the self-made system ( 106) is misaligned with the pitch axis X axis (201) (301, 302), the alignment error is theoretically derived in the coordinate system of the self-made lidar 3D imaging system (713) (301, 302) For the influence of the angle measurement error of the self-made laser radar three-dimensional imaging system (713), an angle measurement error model of the self-made laser radar three-dimensional imaging system (713) is obtained. According to the principle of trigonometric function approximation, the angle measurement error model of the self-made LiDAR 3D imaging system (713) is simplified. According to the principle of error transfer, the angle measurement error model of the self-made LiDAR 3D imaging system (713) is transformed into a Cartesian coordinate system, and the point cloud error model of the self-made LiDAR 3D imaging system (713) is obtained. Use self-made laser radar 3D imaging system (713) and high-precision 3D scanner (714) to scan the same target ball (701-712) to obtain two sets of target point cloud data respectively, and substitute the data into In the point cloud error (expressions (15) and (16) of the manual) of the self-made lidar 3D imaging system (713), solve the model parameters, that is, the alignment error (301, 302), and use this result to realize the error correction of the point cloud of the self-made lidar three-dimensional imaging system (713). Therefore, the specific implementation plan can be divided into four steps: the establishment of the angle measurement error model of the self-made LiDAR 3D imaging system, the simplification of the angle measurement error model of the self-made LiDAR 3D imaging system, and the establishment of the point cloud error model of the self-made LiDAR 3D imaging system And model parameter solution and calibration.

(1) Establishment of angle measurement error model of self-made LiDAR 3D imaging system

The spatial diagram of the target (107) scanned by the self-made laser radar three-dimensional imaging system (713) is shown in Figure 1. When the self-made laser radar three-dimensional imaging system (713) scans, the ideal incident light (105) passes through the optical system (101) Incident to point O in the center of the 45° rotating mirror (103), this point is taken as the coordinate origin. The motor (102) controls the vertical rotation of the 45° rotating mirror (103), the rotation axis of the motor (102) is the pitch axis, which coincides with the X-axis (201), and the positive direction of the X-axis (201) is the same as the incident direction of the ideal incident light (105) . At the same time, the cloud platform (104) rotates horizontally around the vertical axis, the Z axis (203) coincides with the vertical axis, and the positive direction is vertically upward. A right-handed coordinate system is established, and the initial laser emission direction is considered to be the positive direction of the Y axis (202), so as to realize the scanning of the three-dimensional space.

The schematic diagrams of the alignment errors (301, 302) between the incident light and the pitch axis in the coordinate system of the self-made lidar 3D imaging system (713) are shown in Figures 1 and 3. In actual situations, the incident light (106) and the pitch axis There is an angle β (301) between (X-axis) (201), and there is an angle between the opposite vector of the actual incident light (106) projected on the YOZ plane and the azimuth axis (Z-axis) (203) (302), counting from the positive semi-axis of the Z axis, the clockwise direction is positive, and the range is from 0° to 360°. The target object (107) is a target ball, and any point on the target ball is point P (108). It is considered that the outgoing light passes through point P (108) in space. At this time, the azimuth (303) and pitch angle of the outgoing laser are (304) are measured as with The normal vector of the 45 ° rotating mirror (103) under the O-XYZ coordinate system is as follows:

The actual incident light (106) vector is From the vector form of the reflection law, the outgoing light vector can be obtained According to the conversion relationship between spherical coordinates and rectangular coordinates, such as formula (9), the true values of azimuth (303) and pitch angle (304) can be obtained According to the principle of trigonometric function approximation, the azimuth error and pitch error It can be approximated as follows:

The angle measurement error model of the self-made lidar 3D imaging system (713) finally established includes the azimuth error and pitch error The error model results are as follows:

Therefore, the angle measurement error (azimuth angle error) of the self-made laser radar 3D imaging system (713) pitch angle error ) subject to alignment error (301, 302) and azimuth (303), pitch angle (304) influence. Assuming that the self-made laser radar three-dimensional imaging system (713) alignment error (301, 302) parameter is β=5", The vertical scanning range is The horizontal scan range is Angle measurement error (azimuth error pitch angle error ) The simulation results of the change are shown in Fig. 4. The results show that the angle measurement error remains unchanged during the horizontal scanning process; the azimuth error Always increasing, pitch angle error Decrease first and then increase, the order of magnitude of angle measurement error is 10 ^-5 . The simulation results can intuitively show that there is a certain alignment error in the self-made LiDAR 3D imaging system (713) (301, 302), the change of angle measurement error during the whole scanning process.

(2) Simplification of the angle measurement error model of the self-made LiDAR 3D imaging system

The above-mentioned self-made lidar 3D imaging system (713) angle measurement error model is relatively complicated and needs to be simplified reasonably. According to the trigonometric function approximation principle sinβ=β, cosβ=1, and reasonably omit some high-order terms of β, the simplified angle measurement error model of the self-made LiDAR 3D imaging system (713) is obtained as follows:

Assuming that the alignment error (301, 302) parameter is β=5", The vertical scanning range is The horizontal scan range is It can be seen from the analysis that the alignment error (301, 302) caused by the self-made LiDAR 3D imaging system (713) angle measurement error only increases with the pitch angle (304) changes with changes, compare the models before and after simplification, and the comparison results of the simulation before and after the simplification of the self-made lidar three-dimensional imaging system (713) angle measurement error model are shown in Figure 5. The results show that the azimuth error before and after simplification and pitch error with pitch angle (304) The changes of the measured values are the same, the model simplification process is correct, and the simplified model is used as the angle measurement error model of the self-made LiDAR 3D imaging system (713).

(3) Establishment of the point cloud error model of the self-made lidar 3D imaging system

The self-made laser radar three-dimensional imaging system (713) finally obtains and displays the point cloud data of the target target ball (107), and the data format is the three-dimensional coordinates (x, y, z) of the target point (108) in the Cartesian coordinate system ^. (306), the relation of coordinate conversion is as formula (9), and wherein L is the measured value of the one-dimensional distance (305) of target point, is the measured value of target azimuth (303), is the measured value of pitch angle (304).

According to the error transfer formula, the point cloud error model of the self-made lidar 3D imaging system (713) is obtained as follows:

Among them, Δx, Δy, and Δz are the coordinate component error values of the X, Y, and Z axes (201-203) respectively, and ΔL is the system ranging error, which can be regarded as a fixed value within a certain ranging range. is the system azimuth error, is the pitch angle error of the system, which can be expressed by the angle measurement error model of the self-made lidar three-dimensional imaging system (713).

(4) Parameter solution and calibration of the model

The flow diagram of parameter establishment in the point cloud error model of the self-made LiDAR 3D imaging system (713) is shown in Figure 6. N target balls (701-712) are taken as targets, and the self-made LiDAR 3D imaging system (713) Scanning it with a high-precision three-dimensional scanner (714) can obtain point cloud data of multiple points on each target ball (701-712), and use the least square method to analyze the points on each target ball (701-712). Space fitting of the point cloud is carried out, and the coordinates ( _xi ,y _i , _zi ) ^T ,(i=1,2,…,N) and the high-precision three-dimensional Coordinates (x′ _i , y′ _i , z′ _i ) ^T , (i=1, 2, . . . , N) of the center of the sphere under the scanner ( 714 ). Rotate and translate (x′ _i ,y′ _i ,z′ _i ) ^T , (i=1,2,…,N) into the coordinate system of the self-made lidar 3D imaging system (713), and substitute into the self-made laser In the point cloud error expressions (15) and (16) of the radar three-dimensional imaging system (713), the least square method is used to solve the model parameters, and the alignment error is obtained by (301, 302) Realize the point cloud error correction of the self-made lidar three-dimensional imaging system (713). The specific scanning scheme is as follows:

The schematic diagram of the scanning scheme established by the parameters in the point cloud error model of the self-made lidar 3D imaging system (713) is shown in Figure 7. N=12 is taken, that is, 12 target balls (701-712) are prepared, and any Layout, place the self-made laser radar three-dimensional imaging system (713) at the center, make the target ball (701-712) spread over the horizontal scanning range of the self-made laser radar three-dimensional imaging system (713), and spread all over the self-made three-dimensional laser radar imaging system (713) ) within the vertical scanning range, the distance between each target ball (701-712) and the self-made lidar three-dimensional imaging system (713) is roughly the same, and the ranging error is a known constant at this time. Scan all the target spheres (701-712) with the self-made laser radar 3D imaging system (713), and obtain the center coordinates (x _i , y _i , z) of all target spheres (701-712) after center fitting _i ) ^T , (i=1,2,...,12), as the measured value, and obtain the one-dimensional distance L _i (305) and pitch angle of each target ball (701-712) center (304) and azimuth (303). Use a high-precision three-dimensional scanner (714) to perform secondary scanning to obtain the center coordinates (x′ _i , y′ _i , z′ _i ) ^T , (i=1,2,…) of all target balls (701-712) ,12), transform it into the coordinate system of the self-made lidar 3D imaging system (713) through rotation and translation, as the real value, the rotation matrix R and the translation vector T are as follows:

In the formula, a, b, and c are respectively the angles of anticlockwise rotation around each axis of the coordinate system of the high-precision three-dimensional scanner (714), and x ₀ , y ₀ , and z ₀ are three points of X, Y, and Z (201-203) respectively. The displacement in the direction, the point cloud error model of the self-made lidar 3D imaging system (713) can be written as follows:

In the above model, there are a,b,c,x ₀ ,y ₀ ,z ₀ ,β, There are 8 unknown parameters in total. Substituting the center coordinates of 4 target spheres (701, 704, 707, 710) into Equation (19), 12 nonlinear equations can be obtained. Assume first that there are no alignment errors (301, 302), that is, the right side of the equal sign in formula (19) is 0, and the center coordinate data of the target ball (701, 704, 707, 710) are substituted into, and a can be obtained by using the least square method to solve nonlinear equations, Initial values of b,c,x ₀ ,y ₀ ,z ₀ , default alignment error (301, 302) The initial value is 0. After the initial value is determined, the least square method is used to solve 8 unknown parameters, and the alignment error (301, 302) are substituted into formulas (11) and (12) to obtain the system point cloud error (Δx, Δy, Δz) ^T , using the remaining 8 target balls (702, 703, 705, 706, 708, 709, 711 and 712) The coordinate data of the center of the sphere is used to check the model, and the model is continuously corrected. The final model can realize the alignment error of the self-made laser radar 3D imaging system (713) Point cloud error correction of (301, 302), the corrected target point cloud coordinates are (x+Δx, y+Δy, z+Δz) ^T .

To sum up, the present invention proposes a point cloud error correction method based on self-made ground-based laser radar alignment error. The influence of light and pitch axis alignment error on the angle measurement error of the self-made Lidar 3D imaging system, establish a point cloud error model of the self-made Lidar 3D imaging system, and use the point cloud coordinates of the high-precision 3D scanner as the real value to solve the alignment error , so as to realize the error correction of the self-made lidar 3D imaging system for the alignment error target point cloud. The invention is a system point cloud error correction method with strong theory and logic, starting from the error source of the self-made laser radar three-dimensional imaging system, which can be realized by using a small number of target points. A three-dimensional scanning system that uses a mirror and a pan-tilt rotation scanning method.

The above is only the basic scheme of the specific implementation method of the present invention, but the protection scope of the present invention is not limited thereto, and any conceivable change or replacement within the technical scope disclosed by the present invention by anyone familiar with the technical field, All should be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention should be determined by the protection scope of the claims. All changes that come within the equivalent meaning and range of the claims are intended to be included in the scope of the claims.

Claims (8)

1. A point cloud error correction method based on self-made ground-based laser radar alignment error, it is characterized in that, this method is aimed at the scan mode that 45 ° rotating mirror cooperates the pan-tilt rotation in the self-made laser radar three-dimensional imaging system, wherein said self-made laser The radar three-dimensional imaging system includes an optical system, a scanning mechanism (motor, 45° rotating mirror), and a cloud platform; the incident light is emitted to the center (O point) of the 45° rotating mirror through the optical system, and is rotated vertically with the scanning mechanism , the horizontal rotation of the platform is emitted from the self-made laser radar three-dimensional imaging system; the coordinate system (O-XYZ) of the self-made laser radar three-dimensional imaging system includes a pitch axis (X axis), an initial outgoing light direction (Y axis ) and the azimuth axis (Z axis); ideally, the incident light coincides with the pitch axis (X axis), but in reality there is an angle between the incident light and the pitch axis (X axis) β, there is an angle between the opposite vector projected by the incident light on the YOZ plane and the azimuth axis (Z axis) The alignment error is defined as: ① the angle β between the actual incident light and the pitch axis (X axis); ② the opposite vector of the actual incident light projected on the YOZ plane and the azimuth Angle between axes (Z-axis) Counting from the positive semi-axis of the Z axis, the clockwise direction is positive, and the range is 0° to 360°; the one-dimensional distance L of a point P on the target in the coordinate system (O-XYZ) is defined as length, azimuth defined as The angle between the projection on the XOY plane and the X-axis is calculated from the positive semi-axis of the X-axis, and the counterclockwise direction is positive, the range is 0° to 360°, and the pitch angle defined as 90° with and the angle difference between the Z axis, the one-dimensional distance L, the azimuth angle and the pitch angle The measurement errors are ranging error ΔL, azimuth angle error and pitch error The alignment error is analyzed by theoretical For the self-made laser radar three-dimensional imaging system angle measurement error (azimuth angle error pitch angle error ), the angle measurement error model of the self-made lidar three-dimensional imaging system is established; the point cloud of the target is defined as the Cartesian coordinates (x, y, z) ^T of a point on the target, and the point cloud error of the target is defined as the The deviation (Δx, Δy, Δz) ^T between the coordinate measurement value of the target point cloud and the true value, according to the point cloud error (Δx, Δy, Δz) ^T of the target and the angle measurement error (azimuth angle error) of the target pitch angle error ), the point cloud error model of the self-made lidar three-dimensional imaging system is established as follows: Obtain the homemade lidar 3D imaging system for the alignment error The corrected target point cloud coordinates are: (x+Δx,y+Δy,z+Δz) ^T Thereby realizing the alignment error of the self-made lidar 3D imaging system Point cloud error correction; The method mainly includes the following seven steps: 1) Establish the coordinate system (O-XYZ) of the self-made laser radar three-dimensional imaging system; the measured value of any point P on the target in the coordinate system (O-XYZ) is the azimuth angle Pitch angle The true value of any point P on the target in the coordinate system (O-XYZ) is the azimuth Pitch angle said with The difference is the azimuth error said with The difference is the pitch angle error 2) Establish the alignment error by the The angle measurement error model of the self-made laser radar three-dimensional imaging system caused; This model describes the angle measurement error (azimuth angle error) respectively pitch angle error ) and the alignment error and the azimuth of the target Pitch angle The mathematical relationship between them is as follows: 3) Based on the alignment error for the The value of the trigonometric function is approximately transformed, and the angle measurement error model of the self-made lidar 3D imaging system is simplified, as follows: 4) According to the principle of error transmission, establish the point cloud error (Δx, Δy, Δz) ^T of the self-made lidar three-dimensional imaging system and the distance measurement error ΔL, the angle measurement error (azimuth error pitch angle error ) between the mapping relationship, as follows: Wherein, L is the one-dimensional distance of any point on the target in the self-made laser radar three-dimensional imaging system, and ΔL is the measurement error of the one-dimensional distance L of any point on the target; 5) The angle measurement error model of the self-made laser radar three-dimensional imaging system is transformed into a Cartesian coordinate system, and the point cloud error model of the self-made laser radar three-dimensional imaging system is obtained as follows: 6) Using the self-made laser radar three-dimensional imaging system to scan the N targets, and obtain the coordinates ( _xi , y _i , z _i ) ^T of the target in the self-made laser radar three-dimensional imaging system, (i =1, 2,, N), as the measured value; use a high-precision three-dimensional scanner to scan the target twice to obtain the coordinates of the target in the high-precision three-dimensional scanner ( _xi ', y _i ', z _i ') ^T , (i=1,2,…,N), which is transformed into the self-made lidar three-dimensional imaging system as the real value; the difference between the measured value and the real value is The point cloud error (Δx _i , Δy _i , Δz _i ) ^T of the self-made lidar three-dimensional imaging system, (i=1,2,...,N); according to the mapping relationship between the spherical coordinate system and the rectangular coordinate system, Obtain the one-dimensional distance L _i and azimuth of the target from the measured value and pitch angle The ranging error ΔL _i of the target is regarded as a known constant when the one-dimensional distance L _i is within a certain range; 7) Solve the model parameters in the point cloud error model of the self-made lidar three-dimensional imaging system, that is, the alignment error Alignment error of the self-made lidar 3D imaging system The point cloud error of the _self -made _{lidar three} -dimensional imaging system is ^corrected ; distance L _i , azimuth angle Pitch angle and ranging error ΔL _i are substituted into the formula (4), 3*N nonlinear equations are obtained, and the alignment in the self-made lidar three-dimensional imaging system point cloud error model is determined by solving the nonlinear equations error The alignment error Substituting into formula (4), on the basis of the target point cloud coordinates (x, y, z) ^T of the self-made laser radar three-dimensional imaging system, the corrected target point cloud of the self-made laser radar three-dimensional imaging system is obtained Coordinates (x+Δx,y+Δy,z+Δz) ^T . 2. a kind of point cloud error correction method based on self-made ground-based laser radar alignment error according to claim 1, is characterized in that, set up described self-made laser radar three-dimensional imaging system coordinate system (O-XYZ); Described self-made The pitch axis of the lidar three-dimensional imaging system is defined as the motor shaft of the 45° rotating mirror, and the azimuth axis is defined as the shaft of the pan/tilt; ideally, the incident light coincides with the pitch axis; the 45° The center of the reflective surface of the rotating mirror, that is, the intersection point where the ideal incident light is incident on the reflecting surface is the coordinate origin O; coincident with the pitch axis, the positive direction and the ideal incident light are incident on the 45° rotating mirror The coordinate axis with the same direction as above is defined as the X axis; the coordinate axis with the same initial exit direction of the ideal incident light is defined as the Y axis; the coordinate axis coincident with the azimuth axis, and the positive direction is vertically upward is defined as the Z axis. 3. a kind of point cloud error correction method based on self-made ground-based laser radar alignment error according to claim 1, is characterized in that, establishes by described alignment error The angle measurement error model of the self-made lidar three-dimensional imaging system caused by it is simplified; because there is an angle β between the actual incident light and the pitch axis (X axis), the actual incident light is in the YOZ plane There is an angle between the opposite vector of the upward projection and the azimuth axis (Z axis) Counting from the positive semi-axis of the Z axis, the clockwise direction is positive, and the range is 0° to 360°; the establishment process of the angle measurement error model of the self-made laser radar three-dimensional imaging system is as follows: The initial normal unit vector of the 45° rotating mirror (under the O-XYZ coordinate system) is: Ideally, the outgoing ray passes through a point P on the target. At this time, the normal vector of the 45° rotating mirror (under the O-XYZ coordinate system) is: In formula (6) is the measured value of the pitch angle of a point P on the target, is the azimuth measurement value of a point P on the target, are respectively rotated counterclockwise around the X-axis by the angle, rotate the Z axis counterclockwise the The rotation matrix of the angle, the rotation matrix is as follows: Substituting formula (5) and formula (7) into formula (6), the normal vector of the 45 ° rotating mirror under the O-XYZ coordinate system can be obtained as follows: Initially the actual incident light vector is After the pan-tilt rotates, it becomes The vector form of the light reflection law is Find the outgoing light vector as According to the conversion relationship between spherical coordinates and rectangular coordinates, as shown in the following formula (9), the true values of the azimuth angle and the pitch angle can be obtained According to the principle of trigonometric function approximation, the azimuth error The pitch error with the It can be approximated as follows: The obtained angle measurement error model of the self-made lidar three-dimensional imaging system includes the azimuth error and pitch error describes the angular error (azimuth error pitch angle error ) and the alignment error and the azimuth The pitch angle the mathematical relationship between According to the trigonometric function approximation principle sinβ=β, cosβ=1, and reasonably omit some high-order terms of β, the angle measurement error model of the self-made lidar 3D imaging system after simplification is obtained as follows: 4. A kind of point cloud error correction method based on self-made ground-based laser radar alignment error according to claim 1, it is characterized in that, the angular measurement error model in the spherical coordinate system of the self-made laser radar three-dimensional imaging system is converted to the The point cloud error model in the Cartesian coordinate system of the self-made laser radar three-dimensional imaging system; the self-made laser radar three-dimensional imaging system can directly obtain the one-dimensional distance L and the azimuth angle of a point P on the target after scanning with the pitch angle The measured value is converted to the three-dimensional coordinates (x, y, z) ^T of a point P on the target under the Cartesian coordinate system of the self-made laser radar three-dimensional imaging system according to formula (9); Due to the existence of the distance measurement error ΔL and the angle measurement error (azimuth angle error pitch angle error ), the coordinate measurement value of a point P on the target is P'(x+Δx, y+Δy, z+Δz), where (Δx, Δy, Δz) ^T is the point cloud error of the target, according to the error transfer Formula, the point cloud error model of the self-made lidar 3D imaging system is obtained as follows: 5. A kind of point cloud error correction method based on self-made ground-based laser radar alignment error according to claim 1, it is characterized in that, the coordinate conversion of described target in described high-precision three-dimensional scanner is to described self-made laser In the radar three-dimensional imaging system; the alignment error in the point cloud error model of the self-made lidar three-dimensional imaging system Unknown and difficult to measure, so use the self-made lidar 3D imaging system and the point cloud coordinate data in the high-precision 3D scanner to solve the alignment error Realize point cloud error correction; Utilize the self-made laser radar three-dimensional imaging system to scan N described targets, obtain the coordinates (x _i , y _i , z _i ) ^T , (i=1,2,…,N), as measured values; substituting the measured values into formula (9), to obtain the one-dimensional distance L _i and pitch angle of all the targets and azimuth Using the high-precision three-dimensional scanner to scan the same target to obtain the coordinates ( _xi ', y _i ', z _i ') ^T of a point on the target in the coordinate system of the high-precision three-dimensional scanner, (i=1,2,...,N); since the ( _xi ,y _i , _zi ) ^T and the ( _xi ',y _i ', _zi ') ^T are in different coordinate systems, Therefore, the ( _xi ', y _i ', z _i ') ^T is converted to the coordinate system of the self-made laser radar three-dimensional imaging system as a real value; The rotation matrix R and translation vector T of the imaging system are expressed as follows: In formulas (13) and (14), a, b, and c are the counterclockwise rotation angles around each axis of the coordinate system of the high-precision three-dimensional scanner, and x ₀ , y ₀ , and z ₀ are the X, Y, and The displacement in the three directions of Z, the coordinates of the ( _xi ', y _i ', z _i ') ^T in the self-made laser radar three-dimensional imaging system are as follows: 6. a kind of point cloud error correction method based on self-made ground-based lidar alignment error according to claim 1 or 5, is characterized in that, solves the model parameter in the point cloud error model of described self-made lidar three-dimensional imaging system, i.e. the alignment error The point cloud error model of the self-made lidar 3D imaging system can be written as follows: The ranging error ΔL _i is mainly affected by the ranging circuit in the self-made laser radar three-dimensional imaging system, and is a known constant when the one-dimensional distance L _i is within a certain range; the ( _xi , y _i , z _i ) ^T , the ( _xi ', y _i ', _zi ') ^T , the one-dimensional distance L _i , the azimuth and the pitch angle Substituting formula (15) and formula (16), and solving 3*N nonlinear equations based on the least square method, the rotation matrix R and the translation vector T(a, b, c, x ₀ , y ₀ , z ₀ ) with the alignment error There are 8 unknowns in total. 7. A kind of point cloud error correction method based on self-made ground-based laser radar alignment error according to claim 1, it is characterized in that, using the described alignment error of solving gained Improve the point cloud error model of the self-made laser radar three-dimensional imaging system, realize the point cloud error correction of the alignment error of the self-made laser radar three-dimensional imaging system; Substituting formula (11) and formula (12), the point cloud error (Δx, Δy, Δz) ^T of the self-made lidar 3D imaging system is obtained, and the target point cloud coordinates ( On the basis of x, y, z) ^T , obtain the alignment error of the self-made laser radar three-dimensional imaging system The corrected target point cloud coordinates (x+Δx, y+Δy, z+Δz) ^T . 8. A kind of point cloud error correction method based on self-made ground-based laser radar alignment error according to claim 1, is characterized in that, is used for the described self-made laser radar three-dimensional imaging system alignment error point cloud error correction Targets include but are not limited to target balls, plane reflective targets, and other objects that can obtain the spatial coordinates of a point on the target.