Incremental segmentation of point cloud

Patent metadata

Field	Value
Publication number	US-11538168-B2
Application number	US-201817251078-A
Country	US
Kind code	B2
Filing date	Jun 15, 2018
Priority date	Jun 15, 2018
Publication date	Dec 27, 2022
Grant date	Dec 27, 2022

Abstract

A method for segmentation of a point cloud includes receiving a first frame of point cloud from a sensor; segmenting the first frame of point cloud to obtain a first set of point clusters representing a segmentation result for the first frame of point cloud; receiving a second frame of point cloud from the sensor; mapping the first set of point clusters to the second frame of point cloud; determining points within the second frame of point cloud which do not belong to the mapped first set of point clusters; segmenting the points within the second frame of point cloud which do not belong to the mapped first set of point clusters to obtain a second set of point clusters; and generating a segmentation result for the second frame of point cloud by combining the first set of point clusters and the second set of point clusters.

First claim

What is claimed is:

1. A method for segmentation of a point cloud, comprising:
receiving a first frame of point cloud from a sensor;
segmenting the first frame of point cloud using a segmentation algorithm to obtain a first set of point clusters representing a segmentation result for the first frame of point cloud, wherein each of the first set of point clusters is represented using one of a first set of Signatured Gaussian Mixture Models, wherein each Signatured Gaussian Mixture Model comprises a signature representing properties of a respective point cluster;
receiving a second frame of point cloud from the sensor;
mapping the first set of Signatured Gaussian Mixture Models to the second frame of point cloud according to a coordinate transformation based on their coordinate information;
determining points within the second frame of point cloud which do not belong to the mapped first set of Signatured Gaussian Mixture Models;
segmenting the points within the second frame of point cloud which do not belong to the mapped first set of Signatured Gaussian Mixture Models using the segmentation algorithm to obtain a second set of point clusters, wherein each of the second set of point clusters is represented using one of a second set of Signatured Gaussian Mixture Models; and
generating a segmentation result for the second frame of point cloud by combining the first set of Signatured Gaussian Mixture Models and the second set of Signatured Gaussian Mixture Models, wherein the generating comprises:
comparing a signature of each of the first set of signatured Gaussian Mixture Models with a signature of each of the second set of signatured Gaussian Mixture Models; and
merging the first set of signatured Gaussian Mixture Models and the second set of signatured Gaussian Mixture Models based on the comparison.

2. The method according to claim 1 , wherein the segmentation algorithm comprises a region growing algorithm, and wherein a growing criteria of the region growing algorithm is defined based on attributes of points within a point cloud.

3. The method according to claim 2 , wherein the attributes of points within a point cloud comprises at least one of: a distance of two neighboring points, a similarity of normal directions of two neighboring points, or a similarity of a reflection rate of points.

4. The method according to claim 3 , wherein the distance of two neighboring points is less than 0.5 m, wherein the similarity of normal directions of two neighboring points is less than 1 degree, and wherein the similarity of a reflection rate of points is less than 10%.

5. The method according claim 1 , wherein the properties comprise length, width, thickness, orientation, ratio of main axes, reflectivity and/or flatness for the corresponding point cluster.

6. The method according to claim 1 , wherein the sensor comprises at least one of: lidar, stero-camera or time-of-flight camera.

7. A non-transitory machine readable storage medium having instructions stored thereon that, when executed, cause a processor to implement the method according to claim 1 .

8. An apparatus for segmentation of a point cloud, comprising a processor further comprising:
a receiving unit configured to receive a first frame of point cloud from a sensor;
a segmenting unit configured to segment the first frame of point cloud using a segmentation algorithm to obtain a first set of point clusters representing a segmentation result for the first frame of point cloud, wherein each of the first set of point clusters is represented using one of a first set of Signatured Gaussian Mixture Models, wherein each Signatured Gaussian Mixture Model comprises a signature representing properties of a respective point cluster;
the receiving unit further configured to receive a second frame of point cloud from the sensor;
a mapping unit configured to map the first set of Signatured Gaussian Mixture Models to the second frame of point cloud according to a coordinate transformation based on their coordinate information;
a determining unit configured to determine points within the second frame of point cloud which do not belong to the mapped first set of Signatured Gaussian Mixture Models;
the segmenting unit further configured to segment the points within the second frame of point cloud which do not belong to the mapped first set of Signatured Gaussian Mixture Models using the segmentation algorithm to obtain a second set of point clusters, wherein each of the second set of point clusters is represented using one of a second set of Signatured Gaussian Mixture Models; and
a generating unit configured to generate a segmentation result for the second frame of point cloud by combining the first set of Signatured Gaussian Mixture Models and the second set of Signatured Gaussian Mixture Models, wherein the generating unit is configured to generate the segmentation result by:
comparing a signature of each of the first set of signatured Gaussian Mixture Models with a signature of each of the second set of signatured Gaussian Mixture Models; and
merging the first set of signatured Gaussian Mixture Models and the second set of signatured Gaussian Mixture Models based on the comparison.

9. A system for use in a vehicle, comprising:
a sensor; and
an apparatus for segmentation of a point cloud according to claim 8 .

10. A vehicle comprising the system according to claim 9 .

11. An apparatus for segmentation of a point cloud, comprising:
a memory with computer executable instructions stored therein; and
a processor, coupled to the memory and configured to:
receive a first frame of point cloud from a sensor;
segment the first frame of point cloud using a segmentation algorithm to obtain a first set of point clusters representing a segmentation result for the first frame of point cloud, wherein each of the first set of point clusters is represented using one of a first set of Signatured Gaussian Mixture Models, wherein each Signatured Gaussian Mixture Model comprises a signature representing properties of a respective point cluster;
receive a second frame of point cloud from the sensor;
map the first set of Signatured Gaussian Mixture Models to the second frame of point cloud according to a coordinate transformation based on their coordinate information;
determine points within the second frame of point cloud which do not belong to the mapped first set of Signatured Gaussian Mixture Models;
segment the points within the second frame of point cloud which do not belong to the mapped first set of Signatured Gaussian Mixture Models using the segmentation algorithm to obtain a second set of point clusters, wherein each of the second set of point clusters is represented using one of a second set of Signatured Gaussian Mixture Models; and
generate a segmentation result for the second frame of point cloud by combining the first set of Signatured Gaussian Mixture Models and the second set of Signatured Gaussian Mixture Models by:
comparing a signature of each of the first set of signatured Gaussian Mixture Models with a signature of each of the second set of signatured Gaussian Mixture Models; and
merging the first set of signatured Gaussian Mixture Models and the second set of signatured Gaussian Mixture Models based on the comparison.

Assignees

• Bayerische Motoren Werke Ag

Inventors

• Doemling Maximilian
• Jiang Wanli
• Li Qianshan

Classifications

• G06T7/187

  involving region growing; involving region merging; involving connected component labelling · CPC title

• G06T7/11 Primary

  Region-based segmentation · CPC title

• G06T2207/30244

  Camera pose · CPC title

• G06T2207/10028

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

• G06T7/143

  involving probabilistic approaches, e.g. Markov random field [MRF] modelling · CPC title