A method for real-time adjusting the resolution of 3D lidar data

Abstract

The present invention discloses a method for real-time resolution adjustment of 3D lidar data, including: S1: Establish a coordinate system with the center of the lidar as the coordinate origin, where the z-axis is the horizontal scanning rotation axis, the angle between the laser ray and the z-axis is the vertical scanning angle θ, the angle between the projection of the laser ray on the xOy plane and the x-axis is the horizontal scanning angle φ, and the y-axis is the vertical scanning rotation axis when φ = 0; S2: Determine the vertical scanning angle sequence and the horizontal scanning angle sequence; S3: For each scanning direction formed by a pair of scanning angles, obtain the original data segment collected by the lidar near this direction and find the position with the maximum point density in this direction based on the original data segment to obtain the target distance in this direction, thereby obtaining a set of lidar data; S4: Convert each point of all the lidar data obtained in step S3 into a rectangular coordinate system coordinate to obtain a point cloud that meets the required resolution. The present invention can perform upsampling or downsampling on lidar data in real time to change the data resolution.

Description

Technical Field

[0001] The present invention relates to the technical field of lidar resolution adjustment, and more specifically, to a method for real-time adjustment of the resolution of three-dimensional lidar data. Background Art

[0002] A lidar is a laser ranging system, often installed on vehicles such as driverless cars, drones, mobile robots, etc., for constructing environmental maps, detecting targets, etc.

[0003] A three-dimensional lidar (hereinafter referred to as lidar) usually emits laser light from the center to measure the target distance, samples the surrounding three-dimensional space in a scanning manner, and obtains a three-dimensional point cloud (hereinafter referred to as point cloud). The scanning is usually rotated at different rates around two mutually perpendicular axes, so that the sampling points cover the surrounding space of the lidar, and each point of the obtained lidar data is represented by the target distance and the rotation angles of the two axes. This imaging method makes the density of the point cloud of the lidar related to the target distance: the point cloud is sparse at a far distance and dense at a near distance. Moreover, the reflection of light on the object surface will cause ranging errors, so there are a large number of noise points in the point cloud.

[0004] The point cloud is unordered data, so the point cloud density has a great influence on the data processing of the point cloud. When the point cloud is dense, both the processing and display of the data require a large amount of storage space and processing time. Therefore, downsampling is often required first, and downsampling requires searching for neighboring points, which also requires a large amount of space and time. When the point cloud is sparse, the processing result of the data is easily affected by noise points.

[0005] The prior art discloses a device and method for adjustable resolution depth mapping. The device provides a scanning surface and generates a three-dimensional point cloud depicting the depth of the measured surface at each point. In this solution, the device and method utilize a scanning mirror (104) that reflects a laser beam (102) into a scanning line pattern (114). When the raster pattern of the scanning line points to the surface, the reflected laser beam from the surface is received and used to generate a three-dimensional point cloud depicting the depth of the measured surface at each point. The movement of the scanning mirror can be dynamically adjusted to modify the characteristics of the resulting three-dimensional point cloud of the surface. For example, the adjustment of the movement of the scanning mirror can modify the resolution or data density of the resulting three-dimensional point cloud describing the measured surface depth. However, this solution still has the problem of unreasonable point cloud density. Summary of the Invention

[0006] The present invention provides a method for real-time adjustment of the resolution of three-dimensional lidar data, which realizes the adjustment of the resolution of lidar data according to requirements.

[0007] To solve the above technical problems, the technical solution of the present invention is as follows:

[0008] A method for real-time adjustment of the resolution of 3D lidar data, characterized by including the following steps:

[0009] S1: Establish a coordinate system with the lidar center as the coordinate origin. Among them, the z-axis is the horizontal scanning rotation axis, the angle between the laser ray and the z-axis is the vertical scanning angle θ, the angle between the projection of the laser ray on the xOy plane and the x-axis is the horizontal scanning angle φ, and the y-axis is the vertical scanning rotation axis when φ = 0;

[0010] S2: Determine the vertical scanning angle sequence and the horizontal scanning angle sequence

[0011] S3: For each scanning direction formed by the scanning angle pair acquire the original data segment collected by the lidar near this direction and find the position with the maximum point density in this direction based on the original data segment to obtain the target distance in this direction Thus, a set of lidar data is obtained

[0012] S4: Convert each point of all the lidar data obtained in step S3 into the coordinates in the rectangular coordinate system to obtain the point cloud that meets the required resolution

[0013] Preferably, the resolution requirements in step S2 include equal angular intervals or equal spatial intervals.

[0014] Preferably, the equal angular interval is specifically:

[0015] Let and be any starting angles, and there is:

[0016]

[0017]

[0018] In the formula, and are the specified vertical scanning angle interval and horizontal scanning angle interval respectively, which are different from the vertical scanning angle interval δ θ and the horizontal scanning angle interval δ φ .

[0019] Preferably, the equal spatial interval is specifically:

[0020] Let and be any starting angles, and there is:

[0021]

[0022]

[0023] Where η θ and η φ are the average point distances in the specified vertical and horizontal directions respectively, and mean i () represents taking the mean of all distances within the brackets.

[0024] Preferably, in step S3, the original data segment collected by the lidar near this direction is obtained, and the position with the maximum point density in this direction is found based on the original data segment to obtain the target distance Specifically:

[0025] S3.1: Obtain a set of original data segments {(d , θ k , φ k )|k = 1, 2,..., K} collected by the lidar, where the scanning angle satisfies k )|k = 1, 2,..., K}, where the scanning angle satisfies Δ ρ is the range of data for calculating the point density;

[0026] S3.2: Convert each point of the original lidar data segment {(d k , θ k , φ k )|k = 1, 2,..., K} into Cartesian coordinates to obtain the point cloud {x k |k = 1, 2,..., K};

[0027] S3.3: Find the maximum value d k and the minimum value d max and minimum value d min in {d

[0028] |k = 1, 2,..., K}; k |k = 1, 2,..., K}, in the scanning angle pair direction, within the range of [d min , d max , find the d value that maximizes the point density as the target distance in the scanning angle pair direction

[0029] Preferably, in step S3.1, Δ ρ takes a value of 3 to 6max(δ θ , δ φ ).

[0030] Preferably, the point density in step S3.4 Specifically:

[0031]

[0032] In the formula, x(d, θ, φ) is the vector x obtained by converting the point (d, θ, φ) of the lidar data into the coordinates of the rectangular coordinate system, and h is the width of the Gaussian kernel function.

[0033] Preferably, the width h of the Gaussian kernel function in step S3.4 is h = λδd x , where λ = 2 to 6, δ = max(δ θ , δ φ ), and d x is the distance from point x to the origin.

[0034] Preferably, in steps S3.2, S3.4, and S4, the point (d, θ, φ) of the lidar data is converted into the coordinates of the rectangular coordinate system. When there is no assembly error, the conversion formula is:

[0035] x = (x y z) T = d(sinθcosφ sinθsinφ cosθ) T .

[0036] Preferably, in step S3.4, the dichotomy method is used to find the d value that maximizes the point density within the range of [d min , d max . Specifically:

[0037] S3.4.1: Initialization: d 1 = d min , d 2 = d max , λ = 2 to 6, δ = max(δ θ φ , δ φ );

[0038] S3.4.2: Calculate the point density: where h = λδd 1 ; where h = λδd 2 ;

[0039] S3.4.3: If ρ 1 > ρ 2 , then update where h = λδd 2 ; otherwise update where h = λδd 1 ;

[0040] S3.4.4: If d 2-d 1 > ε d then go to step S3.4.3; otherwise, output the maximum value of d as end the search, where ε d is the ranging accuracy of the lidar.

[0041] Compared with the prior art, the beneficial effects of the technical solution of the present invention are as follows:

[0042] The method for real-time resolution adjustment of 3D lidar data according to the present invention can perform upsampling or downsampling on the data in real time when the lidar collects data or processes and displays lidar data, change the data resolution, and can remove the noise points existing in the point cloud. In addition to specifying an increase or decrease in the scanning angle interval, the data resolution can also be changed by specifying the average point distance, so that the point cloud density is relatively uniform and not affected by the target distance. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 is a schematic flow chart of the method of the present invention.

[0044] Figure 2 is the lidar coordinate system provided by the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0045] The drawings are only for illustrative purposes and should not be construed as a limitation of this patent;

[0046] To better illustrate this embodiment, some components in the drawings are omitted, enlarged or reduced, and do not represent the actual size of the product;

[0047] For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0048] The technical solution of the present invention will be further described below with reference to the drawings and embodiments.

[0049] Embodiment 1

[0050] This embodiment provides a method for real-time resolution adjustment of 3D lidar data, as Figure 1 shown, including the following steps:

[0051] S1: Establish a coordinate system with the lidar center as the coordinate origin, as Figure 2 shown, where the z-axis is the horizontal scanning rotation axis, the angle between the laser ray and the z-axis is the vertical scanning angle θ, the angle between the projection of the laser ray on the xOy plane and the x-axis is the horizontal scanning angle φ, and the y-axis is the vertical scanning rotation axis when φ = 0;

[0052] S2: Determine the vertical scanning angle sequence according to the resolution requirement and the horizontal scanning angle sequence

[0053] S3: For each pair of scanning angles For the scanning direction formed, obtain the original data segment collected by the lidar near this direction and find the position with the largest point density in this direction based on the original data segment to obtain the target distance in this direction Thus, a set of lidar data is obtained

[0054] S4: Convert each point of all the lidar data obtained in step S3 into coordinates in a rectangular coordinate system to obtain a point cloud that meets the required resolution

[0055] Embodiment 2

[0056] Based on Embodiment 1, this embodiment further discloses the following content:

[0057] The resolution requirements described in step S2 include equal angular intervals or equal spatial intervals

[0058] The equal angular interval is specifically:

[0059] Let and be any starting angles, and there is:

[0060]

[0061]

[0062] In the formula, and are respectively the specified vertical scanning angle interval and horizontal scanning angle interval, which are different from the vertical scanning angle interval δ θ and the horizontal scanning angle interval δ φ of the lidar device. For example

[0063] The equal spatial interval is specifically:

[0064] Let and be any starting angles, and there is:

[0065]

[0066]

[0067] In the formula, η θ and η φ are respectively the specified average point distance in the vertical direction and the average point distance in the horizontal direction. For example, ηθ = η φ = 5mm, mean i () indicates taking the mean of all distances within the parentheses.

[0068] Example 3

[0069] Based on Example 1 and Example 2, the following content is further disclosed in this example:

[0070] In step S3, find the position with the maximum point density in this direction based on the original data segment to obtain the target distance Specifically:

[0071] S3.1: Obtain a set of original data segments {(d , θ k , φ k ) | k = 1, 2,..., K} collected by a lidar, where the scanning angle satisfies k Δ is the range of data used to calculate the point density; ρ Suppose the lidar is starting to scan from 0°, and the data {(d

[0072] , θ p,q , φ p ) | p = 0, 1,..., P; q = 0, 1,..., Q} has been obtained, where θ q = pδ p , φ θ = qδ p . Find p φ that satisfies 1 p p 2 that satisfies q 1 that satisfies q 2 that satisfies Then the required original data segment is {(d p,q , θ p , φ q ) | p = p 1 , p 1 +1,..., p 2 ; q = q 1 , q 1 +1,..., q 2}}.

[0073] S3.2: The original lidar data segment {(d k , θ k , φ k)|k = 1, 2, ..., K} of each point is converted into Cartesian coordinates to obtain the point cloud {x k |k = 1, 2, ..., K};

[0074] S3.3: Among {d k |k = 1, 2, ..., K}, find the maximum value d max and the minimum value d min ;

[0075] S3.4: According to the point cloud {x k |k = 1, 2, ..., K}, in the scanning angle pair direction, within the range of [d min , d max , find the d value that maximizes the point density as the target distance in the scanning angle pair

[0076] In step S3.1, Δ ρ takes a value of 3 to 6max(δ θ , δ φ ). Specifically, for example, Δ ρ = 3max(δ θ , δ φ ), where δ θ and δ φ are the vertical scanning angle interval and the horizontal scanning angle interval of the lidar device respectively (for example ).

[0077] The point density in step S3.4 is specifically:

[0078]

[0079] In the formula, x(d, θ, φ) is the vector x obtained by converting the point (d, θ, φ) of the lidar data into Cartesian coordinates, and h is the width of the Gaussian kernel function.

[0080] The width h of the Gaussian kernel function in step S3.4 is h = λδd x , where λ = 2 to 6, δ = max(δ θ , δ φ ), and d x is the distance from the point x to the origin.

[0081] In steps S3.2, S3.4, and S4, when converting the point (d, θ, φ) of the lidar data into Cartesian coordinates, the conversion formula without assembly error is:

[0082] x = (x y z)T = d(sinθ cosφ sinθ sinφ cosθ) T 。

[0083] In step S3.4, the dichotomy method is used to find the d value that maximizes the point density within the range of [d min , d max , specifically as follows: The specific steps are as follows:

[0084] S3.4.1: Initialization: d 1 = d min , d 2 = d max , λ = 2 - 6 (for example, λ = 2), δ = max(δ θ , δ φ );

[0085] S3.4.2: Calculate the point density: where h = λδd 1 ; where h = λδd 2 ;

[0086] S3.4.3: If ρ 1 > ρ 2 , then update where h = λδd 2 ; otherwise update where h = λδd 1 ;

[0087] S3.4.4: If d 2 - d 1 > ε d , then go to step S3.4.3; otherwise output the maximum d value as End the search, where ε d is the ranging accuracy of the lidar (for example, 1mm).

[0088] The same or similar reference numerals correspond to the same or similar components;

[0089] The terms used to describe the positional relationship in the drawings are only for illustrative purposes and should not be construed as a limitation of this patent;

[0090] Obviously, the above embodiments of the present invention are merely examples for clearly illustrating the present invention, rather than a limitation on the implementation manners of the present invention. For those of ordinary skill in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation manners here. Any modifications, equivalent replacements, and improvements made within the spirit and principle of the present invention shall be included in the protection scope of the claims of the present invention.

Claims

1. A method for real-time resolution adjustment of 3D lidar data, characterized in that, it includes the following steps: S1: Establish a coordinate system with the lidar center as the coordinate origin. Among them, the z-axis is the horizontal scanning rotation axis, the angle between the laser ray and the z-axis is the vertical scanning angle θ, the angle between the projection of the laser ray on the xOy plane and the x-axis is the horizontal scanning angle φ, and the y-axis is the vertical scanning rotation axis when φ = 0; S2: Determine the vertical scan angle sequence and the horizontal scan angle sequence according to the resolution requirement and the horizontal scan angle sequence S3: For each pair of scanning angles For the scanning direction formed thereby, obtain the original data segment collected by the lidar near this direction and find the position with the highest point density in this direction based on the original data segment to obtain the target distance in this direction Thus, a set of lidar data is obtained S4: Convert each point of all the lidar data obtained in step S3 into Cartesian coordinate system coordinates to obtain a point cloud that meets the required resolution In step S3, obtain the original data segment collected by the lidar near this direction and find the position with the maximum point density in this direction based on the original data segment to obtain the target distance Specifically: S3.1: Obtain a set of original data segments near the scanning direction collected by a lidar, {(d , θ k , φ k )|k = 1, 2, ..., K}, where the scanning angle satisfies k , and Δ ρ is the range of data used to calculate the point density; S3.2: Convert each point of the original lidar data segment {(d k , θ k , φ k )| k = 1, 2, ..., K} into Cartesian coordinates to obtain the point cloud {x k | k = 1, 2, ..., K}; S3.3: Find the maximum value of d k in {d max | k = 1, 2,..., K} and the minimum value of d min ; S3.4: According to the point cloud {x k | k = 1, 2, ..., K}, on the scanning angle pair direction, within the range of [d min , d max , find the d value that maximizes the point density , and use it as the target distance in the scanning angle pair 2. The method for real-time resolution adjustment of 3D lidar data according to claim 1, characterized in that, the resolution requirements in step S2 include equal angular intervals or equal spatial intervals.

3. The method for real-time resolution adjustment of 3D lidar data according to claim 2, characterized in that, the equal angular interval is specifically: Let and be any starting angles, then we have: wherein, and are respectively a specified vertical scanning angle interval and a horizontal scanning angle interval, which are different from the vertical scanning angle interval δ θ and the horizontal scanning angle interval δ φ of the lidar device.

4. The method for real-time resolution adjustment of 3D lidar data according to claim 2, characterized in that, the equal spatial interval is specifically: Let and be any starting angles, then we have: where η θ and η φ are the average point distances in the specified vertical and horizontal directions respectively, and mean i () indicates taking the mean of all distances within the parentheses.

5. The method for real-time resolution adjustment of 3D lidar data according to claim 1, characterized in that, In step S3.1, Δ ρ takes a value of 3 to 6max(δ θ , δ φ ).

6. The method for real-time resolution adjustment of 3D lidar data according to claim 1, characterized in that, Dot density in step S3.4 Specifically: where x(d, θ, φ) is the vector x obtained by converting the point (d, θ, φ) of the lidar data into the rectangular coordinate system coordinates, and h is the width of the Gaussian kernel function.

7. The method for real-time resolution adjustment of 3D lidar data according to claim 6, characterized in that, In step S3.4, the width h of the Gaussian kernel function is h = λδd x , where λ = 2 to 6, δ = max(δ θ , δ φ ), and d x is the distance from point x to the origin.

8. The method for real-time resolution adjustment of 3D lidar data according to claim 7, characterized in that, In steps S3.2, S3.4 and S4, when converting the point (d, θ, φ) of the lidar data into the rectangular coordinate system coordinates, the conversion formula without assembly error is: x = (x y z) T = d(sinθ cosφ sinθ sinφ cosθ) T .

9. The method for real-time resolution adjustment of 3D lidar data according to claim 8, characterized in that, In step S3.4, the dichotomy method is used to find the value of d that maximizes the point density within the range of [d min , d max , specifically as follows: The specific steps are as follows: S3.4.1: Initialization: d 1 = d min , d 2 = d max , λ = 2 to 6, δ = max(δ θ , δ φ ); S3.4.2: Calculate the point density: where h = λδd 1 ; where h = λδd 2 ; S3.4.3: If ρ 1 > ρ 2 , then update where h = λδd 2 ; otherwise update where h = λδd 1 ; S3.4.4: If d 2 - d 1 > ε d , then go to step S3.4.3; otherwise, output the maximum value of d as end the search, where ε d is the ranging accuracy of the lidar.

Images