Airborne LiDAR point cloud filtering method and device in complex scene and storage medium

Abstract

The invention relates to an airborne LiDAR point cloud filtering method and device in a complex scene and a storage medium. The method comprises the steps of grid segmentation and index establishment; taking the lowest point in the grid as an initial seed point, setting a height difference threshold value and a window size, and extracting an initial ground point by using progressive morphological moving window classification filtering; calculating a gradient value by adopting a K-nearest neighbor method, and performing complementary judgment on non-ground points according to the gradient value to obtain complementary judgment ground points and I-type to-be-classified ground points; encrypting a point cloud ground point by using a triangulated irregular network encryption method to obtain encrypted ground point cloud data; and on the basis of the encrypted ground point cloud data, carrying out supplementary judgment on the I-type points to be subjected to ground classification to obtain effective dense ground point cloud data. According to the method, accurate separation of point cloud ground points and non-ground points in a complex scene can be realized, the problem of partial sparse ground points can be solved, and the precision of the ground points in the complex scene is improved.

Description

Technical Field

[0001] The present application relates to the technical field of point cloud data processing, and in particular to an airborne LiDAR point cloud filtering method, device and storage medium in complex scenarios. Background Art

[0002] In recent years, the LiDAR (Light Detection and Ranging) technology has made great progress. LiDAR point clouds are widely used in the fields of digital elevation model (DEM) generation, forest ecosystem survey, 3D building modeling and geological disaster survey. The process of classifying point clouds into ground points and non-ground points, namely point cloud filtering, plays a very important role in the field of point cloud data processing.

[0003] Airborne LiDAR point cloud filtering methods are mainly divided into several categories: slope-based, morphology-based, segmentation-based, surface fitting-based, machine learning, etc. These filtering methods have good filtering effects in scenes with simple and low-complexity terrain structures. However, point cloud filtering is not robust enough in complex scenes (such as complex terrain with large terrain undulations, alternating discontinuities of steep and gentle terrain, and high vegetation coverage). There is a problem of uneven point cloud density in areas with high vegetation coverage, and good filtering effects cannot be achieved in areas with discontinuous terrain or terrain breaks, which affects the accuracy of digital terrain models. Summary of the invention

[0004] In order to solve at least one of the technical problems mentioned above, the present application proposes an airborne LiDAR point cloud filtering method, device and storage medium in complex scenarios.

[0005] According to some embodiments of the present application, a method for filtering airborne LiDAR point clouds in complex scenes is provided, the method comprising:

[0006] Step 1: grid segmentation and index creation;

[0007] Step 2: Take the lowest point in the grid as the initial seed point, set the height difference threshold and window size, and use progressive morphological moving window classification filtering to extract the initial ground point;

[0008] Step 3: Calculate the slope value using the K nearest neighbor method, and make a supplementary judgment on the non-ground points according to the slope value to obtain the supplementary ground points and the I-type ground classification points;

[0009] Step 4: Encrypt the ground points of the point cloud using an irregular triangulated network encryption method to obtain encrypted ground point cloud data;

[0010] Step 5: Based on the encrypted ground point cloud data, the Class I ground points to be classified are re-judged to obtain effective dense ground point cloud data.

[0011] In some possible implementations, the grid segmentation and indexing includes:

[0012] The point cloud data is divided into multiple grids according to the grid size, and each grid stores the index data of the points falling within the grid.

[0013] In some possible implementations, the setting of the height difference threshold and the window size is specifically as follows:

[0014] Height difference threshold dh T,i As shown in formula (1):

[0015]

[0016] In formula (1), dh0 is the initial height difference threshold, s is the average terrain slope, n is the grid size, and dh m is the maximum height difference threshold;

[0017] Window size w i As shown in formula (2):

[0018] w i =2b i +1 (2)

[0019] In formula (2), i is 0, 1, 2, ..., p; p is a positive integer; and b is 2.

[0020] In some possible implementations, calculating the slope value using the K nearest neighbor method includes:

[0021] Select the points to be classified, take each point to be classified as the center, set the radius, and use the K nearest neighbor method to find the nearest N ground points within the radius of each point to be classified, where N ≥ 3;

[0022] Calculate the slope value s of the point to be classified and the adjacent ground point set, and determine the slope domain s of the adjacent ground point set b ;

[0023] Among them, the slope domain s b As shown in formulas (3) and (4):

[0024]

[0025] s b ={s i-N1 ,s i-N2 ,…,s i-N} (4)

[0026] In formulas (3) and (4), i is the point to be classified, N is the total number of adjacent ground points, and X i , Y i , Z i is the three-dimensional coordinate of the point to be classified, X N , Y N , Z N is the three-dimensional coordinate of a nearby ground point, s i-N is the slope value between the point i to be classified and a neighboring ground point, s b is the slope domain of the neighboring ground point set.

[0027] In some possible implementations, the performing supplementary judgment on the non-ground point according to the slope value to obtain the supplementary judged ground point and the Class I ground classification point includes:

[0028] When the slope of the point to be classified and the set of adjacent ground points remain smooth, that is, the slope value is rounded to 0, the point to be classified is determined to be a supplementary ground point; if the point to be classified has one adjacent ground point that is a steep point, then when the point to be classified has at least two adjacent ground points whose slopes remain smooth, the point to be classified is determined to be a supplementary ground point; if the number of adjacent ground points of the point to be classified is less than 3, the point to be classified is marked as a Class I ground point to be classified.

[0029] In some possible implementations, encrypting the point cloud ground points using an irregular triangulated network encryption method to obtain encrypted ground point cloud data includes:

[0030] Merging the initial ground points and the supplementary ground points to obtain point cloud ground points;

[0031] Select the encrypted point to be classified, and calculate the parameters of the encrypted point to be classified; wherein the parameters are the distance d1 and the angle (α, β, γ) from the encrypted point to be classified to the TIN surface threshold of three adjacent ground points;

[0032] If the distance d1 is less than the preset distance threshold dmax, and the angle is less than the preset angle threshold, the to-be-classified encrypted point is determined as a Class I ground encrypted point; with the to-be-classified encrypted point as the center, a circle is added around a preset radius r for circular encryption to obtain a Class I ground encrypted point set;

[0033] If the distance d1 is greater than the distance threshold dmax, a mirror image method is used for determination; wherein the mirror image method is used for determination, including:

[0034] Set the node of the TIN closest to the encrypted point to be classified as the center, and move the encrypted point to be classified symmetrically to the other side of the center to obtain a mirror point;

[0035] Calculate the distance d2 from the mirror point to the TIN surface area. If the distance d2 is less than the distance threshold dmax, determine the corresponding encrypted point to be classified whose distance d1 is greater than the distance threshold dmax as a Class II ground encrypted point; take the encrypted point to be classified as the center, add points around a preset radius r to perform circular encryption, and obtain a Class II ground encrypted point set;

[0036] The initial ground points, the supplementary ground points, the type I ground encrypted point set, and the type II ground encrypted point set are combined to obtain encrypted ground point cloud data.

[0037] In some possible implementations, the performing supplementary judgment on the Class I ground points to be classified based on the encrypted ground point cloud data to obtain effective dense ground point cloud data includes:

[0038] The Class I ground points to be classified are used as new points to be classified, the slope value is calculated by using the K nearest neighbor method based on the encrypted ground point cloud data, and the Class I ground points to be classified are re-judged according to the slope value until a Class I re-judged ground point is obtained;

[0039] The encrypted ground point cloud data is combined with the Class I supplementary ground points to obtain effective dense ground point cloud data.

[0040] According to other embodiments of the present application, an airborne LiDAR point cloud filtering device for complex scenes is provided, including a processor and a memory, wherein at least one instruction or at least one program is stored in the memory, and the at least one instruction or at least one program is loaded and executed by the processor to implement the airborne LiDAR point cloud filtering method for complex scenes as described above.

[0041] According to other embodiments of the present application, a storage medium is provided, in which at least one instruction or at least one program is stored, and the at least one instruction or at least one program is loaded and executed by a processor to implement the airborne LiDAR point cloud filtering method in the complex scene as described above.

[0042] Implementing the embodiments of the present application has the following beneficial effects:

[0043] In view of the fact that the existing point cloud filtering algorithms are not robust enough in complex scenes, combined with the characteristics of uneven point cloud density under high vegetation coverage and complex terrain, the present invention provides an airborne LiDAR point cloud filtering method under complex scenes. By using the step-by-step optimization combination of initial ground point extraction, ground point supplementation, and ground point encryption, not only can the accurate separation of point cloud ground points and non-ground points in complex scenes be achieved, but also the problem of some sparse ground points can be solved. First, the initial ground points are extracted using progressive morphological moving window classification filtering, and then the ground points are supplemented by introducing the slope threshold of the nearest point set. The number of ground points is increased according to the irregular triangulated ground encryption method to solve the problem of the small number of point clouds under high vegetation coverage. By optimizing the point cloud quality, the accuracy of ground points in complex scenes is improved, and then a high-precision digital terrain model is generated to obtain the fine structural features of the ground in complex scenes. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions and advantages in the embodiments of this specification or the prior art, the drawings required for use in the embodiments or the prior art descriptions are briefly introduced below. Obviously, the drawings described below are only some embodiments of this specification. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0045] Figure 1 A flowchart showing a method for filtering airborne LiDAR point clouds in a complex scenario according to an embodiment of the present application is shown;

[0046] Figure 2 A schematic diagram of a progressive morphological moving window according to an embodiment of the present application is shown;

[0047] Figure 3 A schematic diagram showing an irregular triangulated network encryption method for encrypting ground points according to an embodiment of the present application is shown;

[0048] Figure 4 A schematic diagram showing a distance d1 and an angle (α, β, γ) from a to-be-classified encrypted point to a TIN surface threshold of three adjacent ground points according to an embodiment of the present application;

[0049] Figure 5 A schematic diagram showing a method of making a judgment using a mirror image according to an embodiment of the present application is shown;

[0050] Figures 6A-6D A schematic diagram showing filtering results of an airborne LiDAR point cloud filtering method in a complex scene according to an embodiment of the present application is shown. DETAILED DESCRIPTION

[0051] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0052] like Figure 1 As shown, the embodiment of the present application provides an airborne LiDAR point cloud filtering method in a complex scene, including:

[0053] S100: Grid segmentation and index creation.

[0054] In the embodiment of the present application, the grid segmentation and index establishment include:

[0055] The point cloud data is divided into multiple grids according to the grid size, and each grid stores the index data of the points falling within the grid.

[0056] S200: Taking the lowest point in the grid as the initial seed point, setting the height difference threshold and window size, and using progressive morphological moving window classification filtering to extract the initial ground point.

[0057] In the embodiment of the present application, the height difference threshold and the window size are set as follows:

[0058] Height difference threshold dh T,i As shown in formula (1):

[0059]

[0060] In formula (1), dh0 is the initial height difference threshold, s is the average terrain slope, n is the grid size, and dh m is the maximum height difference threshold;

[0061] Window size w i As shown in formula (2):

[0062] w i =2b i +1 (2)

[0063] In formula (2), i is 0, 1, 2, ..., p, where p is a positive integer and b is 2.

[0064] like Figure 2 As shown, in the embodiment of the present application, the rectangular frame formed by the red dotted lines and solid lines on the blue grid shows the progressive morphological moving window.

[0065] S300: Calculate the slope value using the K nearest neighbor method, and make a supplementary judgment on the non-ground points according to the slope value to obtain supplementary ground points and Class I ground points to be classified.

[0066] In the embodiment of the present application, during the initial ground point extraction process, when using a large window for point cloud filtering, it is easy to misjudge the undulating terrain points as non-ground points. To address this problem, it is necessary to further re-judge these non-ground points. Because the ground in complex scenes is undulating, there are ground discontinuities in many areas such as broken terrain, steep slopes, and isolated rocks, so it is necessary to re-judge these unclassified points that are easily misjudged as non-ground points.

[0067] Specifically, the calculation of the slope value using the K nearest neighbor method includes:

[0068] Select the points to be classified, take each point to be classified as the center, set the radius, and use the K-NearestNeighbor (KNN) method to find the N nearest ground points within the radius of each point to be classified, where N ≥ 3;

[0069] Calculate the slope value s of the point to be classified and the adjacent ground point set, and determine the slope domain s of the adjacent ground point set b ;

[0070] Among them, the slope domain s b As shown in formulas (3) and (4):

[0071]

[0072] s b ={s i-N1 ,s i-N2 ,…,s i-N} (4)

[0073] In formulas (3) and (4), i is the point to be classified, N is the total number of adjacent ground points, and X i , Y i , Z i is the three-dimensional coordinate of the point to be classified, X N , Y N , Z N is the three-dimensional coordinate of a nearby ground point, s i-N is the slope value between the point i to be classified and a neighboring ground point, s b is the slope domain of the neighboring ground point set.

[0074] The method of performing supplementary judgment on non-ground points according to the slope value to obtain supplementary judged ground points and Class I ground classification points comprises:

[0075] When the slope of the point to be classified and the set of adjacent ground points remain smooth, that is, the slope value is rounded to 0, the point to be classified is determined to be a supplementary ground point; if the point to be classified has one adjacent ground point that is a steep point, then when the point to be classified has at least two adjacent ground points whose slopes remain smooth, the point to be classified is determined to be a supplementary ground point; if the number of adjacent ground points of the point to be classified is less than 3, the point to be classified is marked as a Class I ground point to be classified.

[0076] S400: Encrypt the ground points of the point cloud using an irregular triangulated network encryption method to obtain encrypted ground point cloud data.

[0077] In the embodiment of the present application, in order to solve the problem of small number and sparse density of ground point clouds under vegetation coverage, the irregular triangulated network encryption method is used to increase the number of ground points, repair point cloud holes, solve the problem of local ground point missing, and improve the accuracy of refined terrain under vegetation coverage. Figure 3 As shown, the red dots represent the points to be classified and encrypted, and the black dots represent the point cloud ground points obtained in the previous steps. The irregular triangulated network is used to encrypt the ground points.

[0078] Specifically, the method of encrypting the ground points of the point cloud using the irregular triangulated network encryption method to obtain the encrypted ground point cloud data includes:

[0079] Merging the initial ground points and the supplementary ground points to obtain point cloud ground points;

[0080] Select the encryption point to be classified and calculate the parameters of the encryption point to be classified; Figure 4 As shown, the parameters are the distance d1 and the angle (α, β, γ) from the encrypted point to be classified to the TIN surface threshold of the three adjacent ground points;

[0081] If the distance d1 is less than the preset distance threshold dmax, and the angle is less than the preset angle threshold, the to-be-classified encrypted point is determined as a Class I ground encrypted point; with the to-be-classified encrypted point as the center, a circle is added around a preset radius r for circular encryption to obtain a Class I ground encrypted point set;

[0082] If the distance d1 is greater than the distance threshold dmax, the mirror method is used for judgment; Figure 5 As shown, the determination using the mirror method includes:

[0083] Set the node of the TIN closest to the encrypted point to be classified as the center, and move the encrypted point to be classified symmetrically to the other side of the center to obtain a mirror point;

[0084] Calculate the distance d2 from the mirror point to the TIN surface area. If the distance d2 is less than the distance threshold dmax, determine the corresponding encrypted point to be classified whose distance d1 is greater than the distance threshold dmax as a Class II ground encrypted point; take the encrypted point to be classified as the center, add points around a preset radius r to perform circular encryption, and obtain a Class II ground encrypted point set;

[0085] The initial ground points, the supplementary ground points, the type I ground encrypted point set, and the type II ground encrypted point set are combined to obtain encrypted ground point cloud data.

[0086] S500: Based on the encrypted ground point cloud data, the Class I ground points to be classified are re-judged to obtain effective dense ground point cloud data.

[0087] In the embodiment of the present application, after the ground point encryption operation is performed, the distribution of the point cloud data changes, and the I-type ground points to be classified are re-judged as ground points based on the new distribution data. The re-judgment operation adopts the same method as S300, that is, the K nearest neighbor method is used to calculate the slope value, and the slope value is used to determine whether the current ground point to be classified is a ground point, so as to complete the re-judgment of the I-type ground points to be classified and obtain the corresponding I-type re-judged ground points.

[0088] Specifically, the method of performing supplementary judgment on the Class I ground points to be classified based on the encrypted ground point cloud data to obtain effective dense ground point cloud data includes:

[0089] The Class I ground points to be classified are used as new points to be classified, the slope value is calculated by using the K nearest neighbor method based on the encrypted ground point cloud data, and the Class I ground points to be classified are re-judged according to the slope value until a Class I re-judged ground point is obtained;

[0090] The encrypted ground point cloud data is combined with the Class I supplementary ground points to obtain effective dense ground point cloud data.

[0091] In the embodiment of the present application, the initial ground points obtained in S200, the supplementary ground points obtained in the first two judgment situations in S300, the Class I ground encrypted point set and Class II ground encrypted point set obtained in S400, and the Class I supplementary ground points obtained by re-supplementing the Class I ground classification points are combined to obtain the final effective dense ground point cloud data. At this step, effective distinction between point cloud ground points and non-ground points can be achieved.

[0092] Combination Figures 6A-6D As shown in Table 1, the measured point cloud data of a certain place is selected in the embodiment of the present application. Fig. 6AThe original ground point cloud map in a complex scene is shown, and the number of original points is 1716905. The above original point cloud data is processed using the airborne LiDAR point cloud filtering method in a complex scene proposed in this application. Figure 6B The initial ground points extracted after processing according to step 2 of the method of the present application are shown, as shown by the orange points, and the number of initial ground points is 33,101. Figure 6C It shows that the supplementary ground points obtained after processing according to step 3 of the method of this application, the initial ground points and the supplementary ground points counted, are counted as the supplementary ground points, which is 39,403. Fig.6D The effective dense ground points obtained after processing according to steps 4 and 5 of the method of this application are shown, and the number is 62472. It can be seen that according to the point cloud filtering method proposed in this application, through the above step-by-step optimization combination method, not only can the accurate separation of point cloud ground points and non-ground points in complex scenes be achieved, but also some sparse ground points can be optimized to achieve effective encryption of sparse ground points.

[0093] Table 1 Point cloud statistics

[0094] Point cloud data quantity Original point 1716905 Initial ground point 33101 Ground point after supplementary judgment 39403 Effective dense ground points 62472

[0095] The method embodiments provided in the embodiments of the present application can be executed in electronic devices such as mobile terminals, computer terminals, servers or similar computing devices.

[0096] The above embodiments have introduced in detail the airborne LiDAR point cloud filtering method in complex scenes. The implementation of the embodiments of the present application has the following beneficial effects:

[0097] The present invention provides an airborne LiDAR point cloud filtering method in complex scenes. Firstly, the initial ground points are extracted by using progressive morphological moving window classification filtering. Then, the ground points are supplemented by introducing the slope threshold of the nearest point set. The number of ground points is increased according to the irregular triangulated ground encryption method to solve the problem of small number of point clouds under high vegetation coverage. The point cloud quality is optimized to improve the accuracy of ground points in complex scenes. Then, a high-precision digital terrain model is generated to obtain the fine structural features of the ground in complex scenes.

[0098] The embodiment of the present application also provides an airborne LiDAR point cloud filtering device in a complex scene, including a processor and a memory, wherein at least one instruction or at least one program is stored in the memory, and at least one instruction or at least one program is loaded and executed by the processor to implement the airborne LiDAR point cloud filtering method in a complex scene as described above. Among them, the memory can be used to store software programs and modules, and the processor executes various functional applications and data processing by running the software programs and modules stored in the memory. In addition, the memory may include a high-speed random access memory and may also include a non-volatile memory. Accordingly, the memory may also include a memory controller to provide the processor with access to the memory. Optionally, the device may also include one or more CPUs or multiple GPUs, one or more power supplies, one or more wired or wireless network interfaces, one or more input and output interfaces 940, and / or, one or more operating systems, and the like.

[0099] An embodiment of the present application also provides a storage medium, in which at least one instruction or at least one program is stored, and the at least one instruction or at least one program is loaded and executed by a processor to implement the airborne LiDAR point cloud filtering method in complex scenarios as described above.

[0100] The embodiments of the present application have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or technical improvements in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for filtering airborne LiDAR point clouds in complex scenes, characterized in that: The method comprises: Step 1: grid segmentation and index creation; Step 2: Take the lowest point in the grid as the initial seed point, set the height difference threshold and window size, and use progressive morphological moving window classification filtering to extract the initial ground point; Step 3: Calculate the slope value using the K nearest neighbor method, and make a supplementary judgment on the non-ground points according to the slope value to obtain the supplementary ground points and the I-type ground classification points; Step 4: Encrypt the ground points of the point cloud using an irregular triangulated network encryption method to obtain encrypted ground point cloud data; Step 5: Based on the encrypted ground point cloud data, the Class I ground points to be classified are re-judged to obtain effective dense ground point cloud data.

2. The method according to claim 1, characterized in that The grid segmentation and index establishment include: The point cloud data is divided into multiple grids according to the grid size, and each grid stores the index data of the points falling within the grid.

3. The method according to claim 1, characterized in that The height difference threshold and window size are specifically set as follows: Height difference threshold dh T,i As shown in formula (1): In formula (1), dh0 is the initial height difference threshold, s is the average terrain slope, n is the grid size, and dh m is the maximum height difference threshold; Window size w i As shown in formula (2): w i =2b i +1 (2) In formula (2), i is 0, 1, 2, ..., p; p is a positive integer; and b is 2.

4. The method according to claim 1, characterized in that The slope value calculation using the K nearest neighbor method includes: Select the points to be classified, take each point to be classified as the center, set the radius, and use the K nearest neighbor method to find the nearest N ground points within the radius of each point to be classified, where N ≥ 3; Calculate the slope value s of the point to be classified and the adjacent ground point set, and determine the slope domain s of the adjacent ground point set b ; Among them, the slope domain s b As shown in formulas (3) and (4): s b ={s i-N1 ,s i-N2 ,…,s i-N } (4) In formulas (3) and (4), i is the point to be classified, N is the total number of adjacent ground points, and X i , Y i , Z i is the three-dimensional coordinate of the point to be classified, X N , Y N , Z N is the three-dimensional coordinate of a nearby ground point, si -N is the slope value between the point i to be classified and a neighboring ground point, s b is the slope domain of the neighboring ground point set.

5. The method according to claim 4, characterized in that The method of performing supplementary judgment on non-ground points according to the slope value to obtain supplementary judged ground points and Class I ground classification points comprises: When the slope of the point to be classified and the set of adjacent ground points remain smooth, that is, the slope value is rounded to 0, the point to be classified is determined to be a supplementary ground point; if the point to be classified has one adjacent ground point that is a steep point, then when the point to be classified has at least two adjacent ground points whose slopes remain smooth, the point to be classified is determined to be a supplementary ground point; if the number of adjacent ground points of the point to be classified is less than 3, the point to be classified is marked as a Class I ground point to be classified.

6. The method according to claim 1, characterized in that The method of encrypting the ground points of the point cloud by using the irregular triangulated network encryption method to obtain the encrypted ground point cloud data comprises: Merging the initial ground points and the supplementary ground points to obtain point cloud ground points; Select the encrypted point to be classified, and calculate the parameters of the encrypted point to be classified; wherein the parameters are the distance d1 and the angle (α, β, γ) from the encrypted point to be classified to the TIN surface threshold of three adjacent ground points; If the distance d1 is less than the preset distance threshold dmax, and the angle is less than the preset angle threshold, the to-be-classified encrypted point is determined as a Class I ground encrypted point; with the to-be-classified encrypted point as the center, a circle is added around a preset radius r for circular encryption to obtain a Class I ground encrypted point set; If the distance d1 is greater than the distance threshold dmax, a mirror image method is used for determination; wherein the mirror image method is used for determination, including: Set the node of the TIN closest to the encrypted point to be classified as the center, and move the encrypted point to be classified symmetrically to the other side of the center to obtain a mirror point; Calculate the distance d2 from the mirror point to the TIN surface area. If the distance d2 is less than the distance threshold dmax, determine the corresponding encrypted point to be classified whose distance d1 is greater than the distance threshold dmax as a Class II ground encrypted point; take the encrypted point to be classified as the center, add points around a preset radius r to perform circular encryption, and obtain a Class II ground encrypted point set; The initial ground points, the supplementary ground points, the type I ground encrypted point set, and the type II ground encrypted point set are combined to obtain encrypted ground point cloud data.

7. The method according to claim 5, characterized in that The performing supplementary judgment on the Class I ground points to be classified based on the encrypted ground point cloud data to obtain effective dense ground point cloud data comprises: The Class I ground points to be classified are used as new points to be classified, the slope value is calculated by using the K nearest neighbor method based on the encrypted ground point cloud data, and the Class I ground points to be classified are re-judged according to the slope value until a Class I re-judged ground point is obtained; The encrypted ground point cloud data is combined with the Class I supplementary ground points to obtain effective dense ground point cloud data.

8. An airborne LiDAR point cloud filtering device in complex scenes, characterized in that: It includes a processor and a memory, wherein at least one instruction or at least one program is stored in the memory, and the at least one instruction or at least one program is loaded and executed by the processor to implement the airborne LiDAR point cloud filtering method in complex scenes as described in any one of claims 1 to 7.

9. A storage medium, characterized in that: The storage medium stores at least one instruction or at least one program, and the at least one instruction or at least one program is loaded and executed by the processor to implement the airborne LiDAR point cloud filtering method in complex scenes as described in any one of claims 1 to 7.

Images