Method and apparatus for point cloud registration, and computer readable medium

Patent metadata

Field	Value
Publication number	US-11158071-B2
Application number	US-202016823288-A
Country	US
Kind code	B2
Filing date	Mar 18, 2020
Priority date	Apr 24, 2019
Publication date	Oct 26, 2021
Grant date	Oct 26, 2021

Abstract

The present disclosure discloses a method and an apparatus for point cloud registration. The method includes: segmenting a source point cloud and a destination point cloud respectively into different categories of attribute features based on semantic; segmenting the source point cloud and the destination point cloud into a plurality of grids based on the attribute features; calculating a current similarity between the source point cloud and the destination point cloud based on the plurality of grids; determining whether the current similarity and a current iterative number satisfy a preset condition; when the current similarity and the current iterative number satisfy the preset condition, performing a registration on the source point cloud and the destination point cloud to obtain a registered result; and based on the registered result, adjusting a position of the source point cloud, updating the current iterative number.

First claim

What is claimed is:

1. A method for point cloud registration, comprising:
segmenting a source point cloud into different categories of attribute features based on semantic, and segmenting a destination point cloud into different categories of the attribute features based on semantic, the attribute features comprising a ground feature and a non-ground feature;
segmenting the source point cloud and the destination point cloud into a plurality of grids based on the attribute features;
calculating a current similarity between the source point cloud and the destination point cloud based on the plurality of grids;
determining whether the current similarity and a current iterative number satisfy a preset condition;
in response to determining that the current similarity and the current iterative number satisfy the preset condition, performing a registration on the source point cloud and the destination point cloud to obtain a registered result; and
based on the registered result, adjusting a position of the source point cloud, updating the current iterative number, and calculating an updated similarity between the destination point cloud and the source point cloud after adjusting the position.

2. The method of claim 1 , wherein determining whether the current similarity and the current iterative number satisfy the preset condition comprises:
comparing the current similarity with a first preset threshold, and comparing the current iterative number with a second preset threshold; and
comparing the current similarity with a previously calculated similarity when the current similarity is smaller than or equal to the first threshold, and the current iterative number is smaller than the second threshold; and
wherein the preset condition comprises: the current similarity being greater than the previously calculated similarity.

3. The method of claim 2 , further comprising:
when the current similarity is greater than the first threshold, or the current iterative number is greater than or equal to the second threshold, decreasing a grid size of each of the plurality of grids according to a preset convergence step size;
determining whether the grid size decreased is greater than a third preset threshold; and
in response to determining that the grid size decreased is greater than the third preset threshold, based on the grid size decreased, segmenting the source point cloud and the destination point cloud into a plurality of grids according to the attribute features.

4. The method of claim 1 , further comprising:
when the current similarity is smaller than or equal to the previously calculated similarity, setting back the position of the source point cloud, and performing the registration on the source point cloud and the destination point cloud.

5. The method of claim 1 , wherein calculating the current similarity between the source point cloud and the destination point cloud based on the plurality of grids comprises:
calculating the current similarity between the source point cloud and the destination point cloud based on the attribute features of the plurality of grids.

6. The method of claim 5 , wherein calculating the current similarity between the source point cloud and the destination point cloud based on the attribute features of the plurality of grids comprises:
for each grid in the source point cloud, selecting a grid closest the grid in the source point cloud from the destination point cloud; and
calculating the current similarity between the source point cloud and the destination point cloud according to an attribute feature of each grid in the source point cloud.

7. The method of claim 1 , wherein performing the registration on the source point cloud and the destination point cloud to obtain the registered result comprises:
selecting a first grid to be matched from the source point cloud based on the attribute features;
determining a second grid to be matched from the destination point cloud according to an attribute feature of the first grid to be matched; and
matching the first grid to be matched with the second grid to be matched.

8. The method of claim 7 , wherein determining the second grid to be matched in the destination point cloud according to the attribute feature of the first grid to be matched comprises:
when the first grid to be matched comprises the ground feature only, selecting grids from the destination point cloud, a distance between each grid selected and the first grid to be matched being smaller than a preset fourth threshold value, calculating a matching degree between each grid selected and the first grid to be matched, and selecting a grid with the highest matching degree as the second grid to be matched; and
when the first grid to be matched comprises the non-ground feature only, or the first grid to be matched comprises the ground feature and the non-ground feature, selecting grids from the destination point cloud, an attribute feature of each grid selected being the same as the attribute feature of the first gird to be matched, and a distance between the grid selected and the first grid to be matched being smaller than the fourth preset threshold, calculating a matching degree between each grid selected and the first grid to be matched, and selecting a grid with the highest matching degree as the second grid to be matched.

9. The method of claim 8 , wherein when the first grid to be matched comprises the ground feature only, calculating the matching degree by a formula of:

S
h

=

e

-

(

h
s

_

-

h
d

_

)

2

Δ
⁢

h

t
⁢
h
⁢
r

2

,

(
1
)

where, h s represents an average height of respective ground features in the first grid to be matched in th

Assignees

• Baidu online network technology beijing co ltd

Inventors

• Yan Miao
• Zhou Zixiang
• Luo Pan
• Bai Yu
• Ma Changjie
• She Dangen

Classifications

• G06T7/33

  using feature-based methods · CPC title

• G06T2207/10028

  Range image; Depth image; 3D point clouds · CPC title

• G06T7/30 Primary

  Determination of transform parameters for the alignment of images, i.e. image registration · CPC title

• G06T2207/30252

  Vehicle exterior; Vicinity of vehicle · CPC title

• G06T7/337 Primary

  involving reference images or patches · CPC title