Method and system for offset correction of UAV surveying and mapping point cloud data based on artificial intelligence

Abstract

The present invention discloses a method and system for offset correction of drone surveying and mapping point cloud data based on artificial intelligence. The method comprises: determining a correction sub-region and its precise surveying and mapping direction according to the building point cloud edge and the surveying and mapping edge in the drone surveying and mapping point cloud data; fitting a point cloud line according to the precise surveying and mapping direction of the correction sub-region, and performing offset correction on the point cloud line according to the drone flight speed. By using the present invention, the accuracy of point cloud data offset correction is improved.

Description

Technical Field

[0001] The present invention relates to the fields of unmanned aerial vehicle surveying and mapping and artificial intelligence technology, and in particular to an artificial intelligence-based unmanned aerial vehicle surveying and mapping point cloud data offset correction method and system. Background Art

[0002] In the prior art, research on point cloud correction is mostly focused on distortion compensation correction, measurement error correction, and system error correction. There is less research on the influence of the motion parameters of the vehicle or aircraft carrying the radar equipment on the point cloud accuracy, and the motion parameters of the carrying equipment also affect the point cloud accuracy. In UAV lidar mapping, the movement of the drone will not only cause point cloud distortion, but also cause the three-dimensional coordinates of the points to be inconsistent with the three-dimensional coordinates of the actual space. Flying too fast or too slow will cause errors in the spatial coordinates of the point cloud. The spatial coordinate error of the point cloud caused by the flight of the drone is called point cloud offset. This offset is particularly evident when the speed of the drone changes. Therefore, it is necessary to correct the coordinate offset of the point cloud data of drone mapping. Summary of the invention

[0003] The purpose of the present invention is to address the defects of the above-mentioned prior art and propose an artificial intelligence-based UAV surveying and mapping point cloud data offset correction method.

[0004] A method for correcting the offset of point cloud data of UAV surveying and mapping based on artificial intelligence, the method comprising:

[0005] Determine the correction sub-area and its precise mapping direction based on the building point cloud edge and mapping edge in the UAV mapping point cloud data;

[0006] The point cloud line is fitted according to the precise mapping direction of the correction sub-area, and the point cloud line is offset corrected according to the flight speed of the UAV.

[0007] Furthermore, the method of determining the correction sub-region and its precise mapping direction according to the building point cloud edge and the mapping edge in the drone mapping point cloud data includes:

[0008] Perform straight line fitting on the surveying edge of the point cloud data to obtain a surveying edge segment set;

[0009] The surveying and mapping sub-areas are divided by using the surveying and mapping edge line segments on one side to obtain a first surveying and mapping sub-area; and the surveying and mapping sub-areas are divided by using the surveying and mapping edge line segments on the other side to obtain a second surveying and mapping sub-area;

[0010] The following operations are performed for the first surveying sub-region and the second surveying sub-region respectively: if the building edge segments of the building point cloud in the surveying sub-region are parallel to each other, the building edge segments that coincide with the minimum circumscribed rectangle of the precise building point cloud are extracted; if the angle between the extended line of the extracted building edge segment and the extended line of the corresponding surveying edge segment meets the preset conditions, the surveying sub-region is used as a candidate correction sub-region;

[0011] The candidate correction sub-region with the larger area is determined as the correction sub-region, and the direction of the surveying edge line segment on one side of the generated correction sub-region is taken as the precise surveying direction.

[0012] Furthermore, the division of the surveying and mapping sub-areas includes: along the rough surveying direction of the UAV, using the endpoints of the surveying and mapping edge line segments to draw perpendicular line segments into the surveying and mapping area, the length of the perpendicular line segment is the surveying width of the UAV, and the closed area composed of the perpendicular line segment and the surveying and mapping edge line segments on both sides is the surveying and mapping sub-area.

[0013] Furthermore, the preset condition is: the angle is between [90°-a, 90°+a] or [-a, a], where a is an angle constant.

[0014] Furthermore, the method also includes: determining a uniform-speed tree point cloud group according to the distribution of center points of the tree point cloud in the UAV mapping point cloud data; if the correction sub-region contains any uniform-speed tree point cloud group, dividing a uniform-speed mapping region from the correction sub-region according to the uniform-speed tree point cloud group, and the offset correction values ​​of all points in the uniform-speed mapping region are the same.

[0015] Furthermore, dividing the uniform speed mapping area from the correction sub-area according to the uniform speed tree point cloud group includes: drawing a straight line perpendicular to the precise mapping direction of the correction sub-area through the center points of the tree point clouds at both ends of the uniform speed tree point cloud group, and the area enclosed by the two straight lines and the mapping edge line segment is the uniform speed mapping area.

[0016] Furthermore, determining the uniform-speed tree point cloud group according to the center point distribution of the tree point cloud in the UAV mapping point cloud data includes: obtaining the center point of the minimum circumscribed rectangle of the tree point cloud in the UAV mapping point cloud data, and determining the tree point cloud group with uniform center point distribution as the uniform-speed tree point cloud group.

[0017] Furthermore, the method for extracting the building point cloud includes: performing threshold analysis on the point cloud elevation data obtained by drone mapping to obtain a first point cloud set whose elevation data is greater than the elevation threshold; clustering the first point cloud set to obtain several subsets, and screening out a rough building point cloud based on the number of points in the subsets and the degree of discreteness of the elevation data; and screening out a precise building point cloud from the rough building point cloud based on the number of echoes and the degree of discreteness of the echo intensity of the lidar data, namely the extracted building point cloud.

[0018] Furthermore, the method for extracting the tree point cloud includes: screening out a rough tree point cloud according to the number of points in the subset and the degree of discreteness of the elevation data; and screening out a precise tree point cloud from the rough tree point cloud according to the number of echoes and the degree of discreteness of the echo intensity of the lidar data, namely the extracted tree point cloud.

[0019] An artificial intelligence-based UAV surveying and mapping point cloud data offset correction system comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the steps of the method are implemented when the computer program is executed by the processor.

[0020] Compared with the prior art, the present invention has the following beneficial effects:

[0021] The present invention uses artificial intelligence analysis technologies such as clustering to analyze point cloud data to achieve correction of point cloud offset, combines building information with surveying and mapping edge information to obtain accurate surveying and mapping directions, and improves the accuracy of point cloud data offset correction. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0023] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0024] The present invention provides an artificial intelligence-based UAV surveying and mapping point cloud data offset correction method, which verifies and implements offset correction based on building point clouds and tree point clouds. Figure 1 The method of the present invention is described below by means of specific embodiments.

[0025] Embodiment 1:

[0026] An artificial intelligence-based UAV surveying and mapping point cloud data offset correction method comprises: determining a correction sub-region and its precise surveying and mapping direction according to the building point cloud edge and the surveying and mapping edge in the UAV surveying and mapping point cloud data; fitting a point cloud line according to the precise surveying and mapping direction of the correction sub-region, and performing offset correction on the point cloud line according to the UAV flight speed.

[0027] In this embodiment, the UAV uses a scanning method to obtain point cloud data for mapping. First, the building point cloud is extracted from the point cloud data obtained by mapping. The method for extracting the building point cloud includes: thresholding the point cloud elevation data obtained by the UAV mapping to obtain a first point cloud set whose elevation data is greater than the elevation threshold; clustering the first point cloud set to obtain a number of subsets, and screening out a rough building point cloud set according to the number of points in the subset and the degree of discreteness of the elevation data; screening the rough building point cloud set according to the number of echoes and the degree of discreteness of the echo intensity of the laser radar data, and the screening obtained The building point cloud set is the extracted building point cloud. It should be noted that the building point cloud set in the present invention refers to a set that includes multiple building point clouds, and one building point cloud represents one building; the same applies to tree point cloud sets.

[0028] Specifically, the point cloud elevation data is first thresholded to obtain the first point cloud set. It should be noted that since the elevation data of the same terrain in different geographical locations may be quite different, the present invention determines the point cloud data within a certain time period as the processing object based on the scanning speed and the flight speed, and assumes that the processing object is the initial point cloud set. In this embodiment, the time period is set to 30 minutes. This step only processes the elevation data of the point cloud. It should be noted that the elevation data of the present invention is calculated and obtained by the TOF ranging principle. The elevation data specifically refers to the z coordinate in the three-dimensional coordinates of the point cloud. Preferably, 0.25 times the mean value of the z coordinates of all points in the initial point cloud set is taken as the threshold, and the threshold processing is performed, wherein the points with z coordinate values ​​lower than the threshold are removed from the initial point cloud set, and the first point cloud set is obtained for subsequent processing. It should be noted that the threshold here can be adjusted according to the actual implementation scenario. The benefits that this step can bring: according to the prior information, that is, the elevation information of terrains such as buildings and trees is higher than that of other terrains, and the first point cloud set is divided based on this, which is conducive to reducing the amount of calculation.

[0029] Then, the first point cloud set is analyzed by the point cloud analysis model to obtain a rough building point cloud set. This step is based on prior information, that is, the terrain corresponding to the extracted first point cloud set usually mainly includes building terrain and tree terrain. The point cloud analysis model is specifically as follows: first, the first point cloud set is clustered based on the clustering method to obtain a number of first point cloud subsets. For each first point cloud subset, the mean square error σ1(z) of the z coordinates of the points in the set is calculated. The purpose of using the mean square error is to represent the degree of discreteness; the maximum value Max(z) of the z coordinates of the points in the set is taken, and the number of points N1 in the interval [Max(z)-d1,Max(z)] is counted. When σ1(z)≤M1 and N1≥M2, the first point cloud subset is judged to be a rough building point cloud set. The reason why the point cloud analysis model uses two parameters is that the mean square error is used to mainly distinguish between building and forest terrain. For non-building buildings, their roof structures do not meet the flatness condition. Therefore, M1 should be set slightly larger to ensure that the point cloud set of this type of building is correctly divided. Preferably, it is set to 3 in this embodiment. If M1 is set slightly larger, some individual tree point clouds will be classified as building point clouds. Therefore, the purpose of setting M2 is to eliminate such errors. Preferably, M2 is 100. The purpose of setting d1 is to prevent misjudgment caused by the special structure of the building roof. Preferably, d1 is set to 1.5m in this embodiment. It should be noted that the implementer can adjust the above parameters appropriately according to actual conditions.

[0030] Furthermore, according to the echo number mul of the rough building point cloud n , echo intensity variance σ2(ei n ) determines the accurate building point cloud, and the obtained accurate building point cloud set is the extracted building point cloud set. Preferably, when mul n =1 and σ2(ei n )≤M3, confirm that the point cloud set category is a building category, that is, determine it as a precise building point cloud set. This step is based on prior information, that is, the number of echoes of buildings is usually 1, while the number of echoes of trees is usually greater than 1, and the echo intensity distribution of buildings is relatively stable, while the echo intensity distribution of trees is relatively discrete. Preferably, in this embodiment, the echo intensity adopts 8 bits and M3 is 6.

[0031] Next, the correction sub-area and its precise mapping direction are determined based on the building point cloud edge and the mapping edge in the UAV mapping point cloud data.

[0032] To perform offset correction, it is necessary to first determine the precise mapping direction. Only by determining the precise mapping direction can we determine the direction in which the point cloud should be corrected. During the drone mapping process, due to the influence of the environment, terrain, etc., the point cloud data is incomplete or the accuracy is reduced, resulting in unreliable edge of the point cloud mapping. In order to ensure the credibility of the edge of the point cloud mapping, the present invention verifies the edge of the point cloud mapping in combination with the building point cloud.

[0033] Specifically, a straight line fitting is performed on the surveying edge of the point cloud data to obtain a set of surveying edge line segments. When fitting a straight line, it is necessary to set the corresponding number of internal points to balance the accuracy of the fitted straight line and the number of fitted straight lines. The surveying edge line segments on one side are used to divide the surveying sub-area to obtain the first surveying sub-area; the surveying edge line segments on the other side are used to divide the surveying sub-area to obtain the second surveying sub-area. The method of dividing the surveying sub-area is: along the rough surveying direction of the UAV, the endpoints of the surveying edge line segments are used to make vertical line segments into the surveying area. The length of the vertical line segment is the surveying width of the UAV. The closed area composed of the vertical line segment and the surveying edge line segments on both sides is the surveying sub-area.

[0034] The following operations are performed for the first surveying sub-region and the second surveying sub-region respectively: if the building edge segments of the building point cloud in the surveying sub-region are parallel to each other, then the building edge segments that coincide with the minimum circumscribed rectangle of the precise building point cloud are extracted; if the angle between the extended line of the extracted building edge segment and the extended line of the corresponding surveying edge segment meets the preset conditions, then the surveying sub-region is used as an alternative correction sub-region; wherein the preset conditions are: the angle is between [90°-a, 90°+a] or [-a, a], and a is an angle constant. The candidate correction sub-region with the larger area is determined as the correction sub-region, and the direction of the surveying edge segment on one side of the generated correction sub-region is used as the precise surveying direction. Regarding the angle constant a, the implementer can adjust it according to the drone scanning parameters. At the same time, in order to improve the judgment accuracy of the precise surveying direction and facilitate implementation, a preferred implementation method is given here, and the angle constant a satisfies the following conditions: Among them, a0 is the initial value of the angle, b0 is the reference value of the angle variable, ρ0 is the reference value of the average scanning angle interval, FOV is the field of view angle range, f0 is the field of view angle reference value, ρ is the average scanning angle interval, a0, b0, f0, ρ0, α, β are constant parameters, preferably, 3, 3, 55, 8, 1 / 3, 2 / 3 respectively. The above angle constant conditions generate angle thresholds under different scanning parameters based on the field of view angle and the average scanning angle interval, which improves the screening accuracy of the candidate correction sub-areas and the accuracy of the precise mapping direction, and further improves the accuracy of the point cloud offset correction.

[0035] Finally, the point cloud line is fitted according to the precise mapping direction of the correction sub-area, the UAV flight speed is determined according to the distance between adjacent point cloud lines, and the point cloud data is offset corrected according to the UAV flight speed.

[0036] After obtaining the precise mapping direction of each correction sub-area, it is first necessary to fit the point cloud row straight line according to the precise mapping direction, extract the points belonging to a row, and then correct the points of the extracted row according to the offset correction value corresponding to the correction sub-area. For a specific correction sub-area, when fitting the point cloud row straight line, the direction of the point cloud row straight line in the correction sub-area is perpendicular to the precise mapping direction, and the fitting is performed using the RANSAC method. The number of point cloud row straight lines that need to be fitted is the same as the number of drone scans.

[0037] In the present invention, a reference coordinate system is established based on the initial position of the drone, and the drone transforms the point cloud collected in real time to the reference coordinate system through projection transformation during flight. Without considering the influence of elevation, the point cloud coordinate deviation of the drone in uniform flight mode is basically the same. When the speed of the drone changes, the point cloud coordinate deviation will change. Therefore, it is also necessary to determine the flight speed of the drone for offset correction. One implementation method is to provide the transformation of the drone speed through an inertial measurement unit and a differential GPS. However, due to the influence of equipment accuracy and sampling period, the accuracy of the obtained speed change is not high. Another implementation method is to directly analyze the speed change of the drone based on the attributes of the point cloud data.

[0038] For a correction sub-region, obtain the distances d1 and d2 between a point cloud row and its two adjacent rows. Assuming the sampling interval is t, the velocity representation value

[0039] The speed characterization value is mapped to the actual flight speed according to the mapping function. The actual flight speed is obtained through the IMU unit. It should be noted that the IMU data is only needed when fitting the function. The speed mapping relationship can be achieved through a mapping table, specifically by statistically analyzing the data pairs of the speed characterization value and the actual flight speed. The speed mapping relationship can also be curve-fitted through multiple sets of speed data, and the least squares method is used to obtain the mapping function v′=f(v). It should be noted that the point cloud row spacing is not only related to the flight speed, but also to the flight altitude. In this embodiment, the aforementioned mapping function and the data pairs in the mapping table all refer to the mapping relationship at the same altitude.

[0040] Different flight speeds correspond to different offset correction values. The points in the point cloud line are offset corrected according to the offset correction value corresponding to the flight speed. The correspondence between the offset correction value and the speed can be realized through a mapping table or a mapping function. According to the flight speed and the offset correction value, the fitting function l = f (v′, z) is performed. Specifically, the least squares method is used to fit multiple sets of data, and the obtained l is the offset correction value. Furthermore, when the drone is flying, it may have an offset in the flight direction and perpendicular to the flight direction. Therefore, it is necessary to fit the mapping functions in the two directions respectively to obtain the corresponding offset correction value. In addition, the offset correction value is affected by the height of the detected object. According to the flight speed, the object elevation value and the offset correction value, the fitting function l = f (v′, z) is performed. The least squares method is also used to fit multiple sets of data, and z is the elevation value. After the correction value is obtained, the points in the point cloud line are offset corrected according to their correction values.

[0041] For points that are not divided into the correction sub-area, the two angles between the precise mapping directions of the two adjacent correction sub-areas and the x-axis direction of the reference coordinate system can be obtained, and the average of the two angles can be obtained. The slope of the straight line corresponding to the average angle is set to k1, and the obtained Use k2 as the reference slope to fit the point cloud line between the two correction areas and calculate the velocity characterization value.

[0042] Embodiment 2:

[0043] This embodiment provides a method for correcting the offset of point cloud data for UAV surveying and mapping based on artificial intelligence. Based on Embodiment 1, this embodiment uses tree point cloud data to improve the efficiency of point cloud offset correction. In urban terrain surveying and mapping, the trees on both sides of the road are similar in shape and size and are evenly distributed. Therefore, the present invention preliminarily determines the movement mode of the UAV based on the point cloud of the trees on both sides of the road, thereby determining the uniform speed surveying area.

[0044] An artificial intelligence-based offset correction method for drone mapping point cloud data also includes: determining a uniform-speed tree point cloud group based on the distribution of center points of tree point clouds in drone mapping point cloud data; if a correction sub-region contains any uniform-speed tree point cloud group, a uniform-speed mapping region is divided from the correction sub-region based on the uniform-speed tree point cloud group, and the offset correction values ​​of all points in the uniform-speed mapping region are the same. Determining a uniform-speed tree point cloud group based on the distribution of center points of tree point clouds in drone mapping point cloud data includes: obtaining the center point of the minimum circumscribed rectangle of the tree point cloud in the drone mapping point cloud data, and determining the tree point cloud group with uniformly distributed center points as the uniform-speed tree point cloud group. Specifically, obtaining the coordinates of the intersection points of the diagonals of each circumscribed rectangle It is the center point of the circumscribed rectangle of the tree point cloud, and k is the kth tree point cloud. When there is no outlier in the distance sequence composed of the distances of adjacent center points (that is, the distances between adjacent center points are basically the same), it is judged that the UAV is in a uniform flight mode and is basically in a uniform flight state in the tree point cloud area. The uniform speed judgment here is based on the prior information that when the elevation effect is not considered, in the uniform flight mode, the offset of the point cloud coordinates is basically the same, so the distance between the adjacent coordinates obtained is still the same, and the same correction value can be used for correction.

[0045] The uniform speed surveying area is divided from the correction sub-area according to the uniform speed tree point cloud group, including: a straight line perpendicular to the precise surveying direction of the correction sub-area is drawn through the center points of the tree point clouds at both ends of the uniform speed tree point cloud group, and the area enclosed by the two straight lines and the surveying edge line segment is the uniform speed surveying area.

[0046] The method for extracting tree point clouds is as follows: First, for the first point cloud subset that does not satisfy σ1(z)≤M1 and N1≥M2, the point cloud set on both sides of the road is screened out through the terrain as a rough tree point cloud set. Next, the rough tree point cloud set category is verified with the category obtained by the lidar data to ensure the accuracy of the point cloud set category. The purpose of this step is to verify the rough tree point cloud set category with the lidar data to ensure the accuracy of the point cloud set category, which can bring the following benefits: the results obtained by the above point cloud analysis model still inevitably have certain errors, and there are also certain errors in the terrain category analysis through echo data. By analyzing the terrain category from different attributes, the method of mutual verification is used to obtain a more accurate point cloud set category to avoid large errors in the subsequent point cloud offset correction. The specific verification method is: for any rough tree point cloud, determine the number of echoes and echo intensity of each point in the point cloud, and count the number of echoes mul n and the mean square error of the echo intensity σ2(ei n ), where n represents the nth point cloud set, when mul n >1 and σ2(ei n )>M3, the point cloud category is the tree category; the verification method is based on prior information, that is, the number of building echoes is usually 1, while the number of tree echoes is usually greater than 1, and the building echo intensity distribution is relatively stable, while the tree echo intensity distribution is relatively discrete. In the present invention, the echo intensity uses 8 bits, and M3 is taken as 6. If the rough tree point cloud category is consistent with the laser radar echo data verification category, the rough tree point cloud category is used as the extracted tree point cloud.

[0047] Embodiment 3:

[0048] This embodiment provides a system embodiment. A system for correcting the offset of point cloud data for UAV surveying and mapping based on artificial intelligence, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the computer program is executed by the processor, the steps of the method for correcting the offset of point cloud data for UAV surveying and mapping based on artificial intelligence are implemented.

[0049] The above embodiments are only preferred embodiments of the present invention and are not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A method for correcting the offset of point cloud data of UAV surveying and mapping based on artificial intelligence, characterized in that: The method includes: Determine the correction sub-area and its precise mapping direction based on the building point cloud edge and mapping edge in the UAV mapping point cloud data; Fit the point cloud line according to the precise mapping direction of the correction sub-area, and perform offset correction on the point cloud line according to the flight speed of the drone; According to the distribution of the center points of the tree point clouds in the UAV surveying point cloud data, the uniform speed tree point cloud group is determined; If the correction sub-region contains any uniform-speed tree point cloud group, a uniform-speed surveying region is divided from the correction sub-region according to the uniform-speed tree point cloud group, and the offset correction values ​​of all points in the uniform-speed surveying region are the same; Determining the correction sub-area and its precise mapping direction according to the building point cloud edge and the mapping edge in the UAV mapping point cloud data includes: Perform straight line fitting on the surveying edge of the point cloud data to obtain a surveying edge segment set; The surveying and mapping sub-areas are divided by using the surveying and mapping edge line segments on one side to obtain a first surveying and mapping sub-area; and the surveying and mapping sub-areas are divided by using the surveying and mapping edge line segments on the other side to obtain a second surveying and mapping sub-area; The following operations are performed for the first surveying sub-region and the second surveying sub-region respectively: if the building edge segments of the building point cloud in the surveying sub-region are parallel to each other, the building edge segments that coincide with the minimum circumscribed rectangle of the precise building point cloud are extracted; if the angle between the extended line of the extracted building edge segment and the extended line of the corresponding surveying edge segment meets the preset conditions, the surveying sub-region is used as a candidate correction sub-region; Determine the candidate correction sub-region with the larger area as the correction sub-region, and use the direction of the surveying edge line segment on one side of the generated correction sub-region as the precise surveying direction; The preset condition is: the angle is [90°-a, 90°+a] or [-a, a], where a is an angle constant; The angle constant a satisfies the conditions: Among them, a0 is the initial value of the angle, b0 is the reference value of the angle variable, ρ0 is the reference value of the average scanning angle interval, FOV is the field of view angle range, f0 is the field of view angle reference value, ρ is the average scanning angle interval, a0, b0, f0, ρ0, α, β are constant parameters; The method for extracting building point clouds comprises: performing threshold analysis on point cloud elevation data obtained by drone mapping to obtain a first point cloud set whose elevation data is greater than an elevation threshold; clustering the first point cloud set to obtain a plurality of subsets, and screening out a rough building point cloud according to the number of points in the subsets and the degree of dispersion of the elevation data; screening out a precise building point cloud from the rough building point cloud according to the number of echoes and the degree of dispersion of the echo intensity of the laser radar data, namely the extracted building point cloud; The tree point cloud extraction method includes: selecting a rough tree point cloud according to the number of points in the subset and the degree of discreteness of elevation data; selecting a precise tree point cloud from the rough tree point cloud according to the number of echoes and the degree of discreteness of echo intensity of laser radar data, that is, the extracted tree point cloud.

2. The method according to claim 1, characterized in that The division of the surveying and mapping sub-areas comprises: Along the rough mapping direction of the UAV, use the endpoints of the mapping edge line segment to make a vertical line segment into the mapping area. The length of the vertical line segment is the mapping width of the UAV. The closed area composed of the vertical line segment and the mapping edge line segments on both sides is the mapping sub-area.

3. The method according to claim 1, characterized in that The step of dividing the uniform speed surveying area from the correction sub-area according to the uniform speed tree point cloud group comprises: A straight line perpendicular to the precise mapping direction of the correction sub-area is drawn through the center points of the tree point clouds at both ends of the uniform speed tree point cloud group. The area enclosed by the two straight lines and the mapping edge line segment is the uniform speed mapping area.

4. The method according to claim 1, characterized in that Determining the uniform speed tree point cloud group according to the center point distribution of the tree point cloud in the UAV surveying point cloud data includes: The center point of the minimum circumscribed rectangle of the tree point cloud in the UAV mapping point cloud data is obtained, and the tree point cloud group with uniform center point distribution is determined as the uniform speed tree point cloud group.

5. An artificial intelligence-based UAV surveying and mapping point cloud data offset correction system, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images