Granted Patent B2

US 12,412,286 · App. 17/973,871 · Granted Sep 9, 2025

Inventors: Qingyuan Jia (Beijing, CN); Junjie Cai (Beijing, CN); Kai Zhong (Beijing, CN); Jianzhong Yang (Beijing, CN); Tongbin Zhang (Beijing, CN); Zhen Lu (Beijing, CN)

Assignee: BEIJING BAIDU NETCOM SCIENCE TECHNOLOGY CO., LTD.

G06T7/30G06T7/10G06T2207/10028

Abstract

Claims (80)

1. A point cloud registration method, comprising:

acquiring target point cloud data;

dividing the target point cloud data into a plurality of point cloud sets;

determining a coincidence degree between every two point cloud sets and determining a fixed point cloud set and a registration point cloud set from two point cloud sets with a coincidence degree between the two point cloud sets being greater than a preset threshold;

determining a target registration matrix between the fixed point cloud set and the registration point cloud set; and

performing registration of the fixed point cloud set with the registration point cloud set according to the target registration matrix;

wherein the determining the target registration matrix between the fixed point cloud set and the registration point cloud set comprises:

determining an initial registration matrix between the fixed point cloud set and the registration point cloud set; and

adjusting the initial registration matrix to determine the target registration matrix;

wherein the initial registration matrix comprises a translation vector; and

adjusting the initial registration matrix to determine the target registration matrix comprises:

determining a smoothing distance based on the translation vector and a preset distance threshold;

determining a weight based on a distance between a point cloud of a coincident part between the fixed point cloud set and the registration point cloud set and a point cloud of a remaining part of the fixed point cloud set and the registration point cloud set, the smoothing distance, and a travelling distance corresponding to the fixed point cloud set; and

determining the target registration matrix based on the weight and the initial registration matrix.

2. The method of claim 1 , wherein determining the initial registration matrix between the fixed point cloud set and the registration point cloud set comprises:

using the closest points between the fixed point cloud set and the registration point cloud set as corresponding point pairs; and

performing iteration by using an average of space distance squares of the corresponding point pairs as an error, and determining the initial registration matrix in response to the error satisfying a preset condition.

3. The method of claim 1 , wherein the initial registration matrix comprises a rotation matrix; and

adjusting the initial registration matrix to determine the target registration matrix comprises:

determining, based on the rotation matrix, Euler-angle expressions corresponding to three coordinate axes of a coordinate system in which the target point cloud data is located; and

adjusting at least one Euler-angle expression, and determining the target registration matrix according to the adjusted Euler-angle expression.

4. The method of claim 1 , wherein determining a coincidence degree between every two point cloud sets comprises:

determining an enclosing frame of each point cloud set; and

determining a coincidence degree between the two point cloud sets according to a coincidence degree between enclosing frames of the two point cloud sets.

5. The method of claim 4 , wherein determining the enclosing frame of each point cloud set comprises:

performing a filtering operation on the point cloud set to determine an enclosing frame of a filtered point cloud set.

6. The method of claim 1 , wherein the determining a registration point cloud set from two point cloud sets with a coincidence degree between the two point cloud sets being greater than a preset threshold comprises:

acquiring a precision of each point cloud set; and

for two point cloud sets with a coincidence degree being greater than the preset threshold, using a point cloud set with a higher precision in the two point cloud sets as a fixed point cloud set, and a point cloud set with a lower precision in the two point cloud sets as a registration point cloud set.

7. The method of claim 1 , wherein dividing the target point cloud data into a plurality of point cloud sets comprises:

dividing the target point cloud data according to a preset time period to obtain the plurality of point cloud sets.

8. The method of claim 1 , wherein the method further comprises:

generating a high-definition map based on the registered point cloud data.

9. A point cloud registration apparatus includes:

at least one processor; and

a memory storing instructions, wherein the instructions when executed by the at least one processor, cause the at least one processor to perform operations, the operations comprising:

acquiring target point cloud data;

dividing the target point cloud data into a plurality of point cloud sets;

determining a coincidence degree between every two point cloud sets and determining a fixed point cloud set and a registration point cloud set from two point cloud sets with a coincidence degree between the two point cloud sets being greater than a preset threshold;

determining a target registration matrix between the fixed point cloud set and the registration point cloud set; and

performing registration of the fixed point cloud set with the registration point cloud set according to the target registration matrix;

wherein the determining the target registration matrix between the fixed point cloud set and the registration point cloud set comprises:

determining an initial registration matrix between the fixed point cloud set and the registration point cloud set; and

adjusting the initial registration matrix to determine the target registration matrix;

wherein the initial registration matrix comprises a translation vector; and

adjusting the initial registration matrix to determine the target registration matrix comprises:

determining a smoothing distance based on the translation vector and a preset distance threshold;

determining a weight based on a distance between a point cloud of a coincident part between the fixed point cloud set and the registration point cloud set and a point cloud of a remaining part of the fixed point cloud set and the registration point cloud set, the smoothing distance, and a travelling distance corresponding to the fixed point cloud set; and

determining the target registration matrix based on the weight and the initial registration matrix.

10. The apparatus of claim 9 , wherein determining the initial registration matrix between the fixed point cloud set and the registration point cloud set comprises:

using the closest points between the fixed point cloud set and the registration point cloud set as corresponding point pairs; and

performing iteration by using an average of space distance squares of the corresponding point pairs as an error, and determining the initial registration matrix in response to the error satisfying a preset condition.

11. The apparatus of claim 9 , wherein the initial registration matrix comprises a rotation matrix; and

adjusting the initial registration matrix to determine the target registration matrix comprises:

determining, based on the rotation matrix, Euler-angle expressions corresponding to three coordinate axes of a coordinate system in which the target point cloud data is located; and

adjusting at least one Euler-angle expression, and determining the target registration matrix according to the adjusted Euler-angle expression.

12. The apparatus according to claim 9 , wherein determining a coincidence degree between every two point cloud sets comprises:

determining an enclosing frame of each point cloud set; and

determining a coincidence degree between the two point cloud sets according to a coincidence degree between enclosing frames of the two point cloud sets.

13. The apparatus according to claim 12 , wherein determining the enclosing frame of each point cloud set comprises:

performing a filtering operation on the point cloud set to determine an enclosing frame of a filtered point cloud set.

14. The apparatus according to claim 9 , wherein the determining a registration point cloud set from two point cloud sets with a coincidence degree between the two point cloud sets being greater than a preset threshold comprises:

acquiring a precision of each point cloud set; and

for two point cloud sets with a coincidence degree being greater than the preset threshold, using a point cloud set with a higher precision in the two point cloud sets as a fixed point cloud set, and a point cloud set with a lower precision in the two point cloud sets as a registration point cloud set.

15. The apparatus of claim 9 , wherein dividing the target point cloud data into a plurality of point cloud sets comprises:

dividing the target point cloud data according to a preset time period to obtain the plurality of point cloud sets.

16. A non-transitory computer readable storage medium storing computer instructions, wherein the computer instructions are used for causing the computer to execute operations comprising:

acquiring target point cloud data;

dividing the target point cloud data into a plurality of point cloud sets;

determining a coincidence degree between every two point cloud sets and determining a fixed point cloud set and a registration point cloud set from two point cloud sets with a coincidence degree between the two point cloud sets being greater than a preset threshold;

determining a target registration matrix between the fixed point cloud set and the registration point cloud set; and

performing registration of the fixed point cloud set with the registration point cloud set according to the target registration matrix;

wherein the determining the target registration matrix between the fixed point cloud set and the registration point cloud set comprises:

determining an initial registration matrix between the fixed point cloud set and the registration point cloud set; and

adjusting the initial registration matrix to determine the target registration matrix;

wherein the initial registration matrix comprises a translation vector; and

adjusting the initial registration matrix to determine the target registration matrix comprises:

determining a smoothing distance based on the translation vector and a preset distance threshold;

determining a weight based on a distance between a point cloud of a coincident part between the fixed point cloud set and the registration point cloud set and a point cloud of a remaining part of the fixed point cloud set and the registration point cloud set, the smoothing distance, and a travelling distance corresponding to the fixed point cloud set; and

determining the target registration matrix based on the weight and the initial registration matrix.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (22)