SYSTEM AND METHOD OF EXTRACTING PLANE FEATURES

• US 20100074473A1
• Filed: 09/19/2008
• Published: 03/25/2010
• Est. Priority Date: 09/19/2008
• Status: Active Grant

• Current Assignee: Honeywell International Inc.
• Original Assignee: Honeywell International Inc.
• Inventors: Kotaba, Ondrej
• Granted Patent: US 8,503,730 B2
• Time in Patent Office: Days
• US Class Current: 382/106
• CPC Class Codes: G06T 2207/10028 Range image; Depth image; 3D point clouds
• CPC Class Codes: G06T 2207/10032 Satellite or aerial image; Remote sensing
• CPC Class Codes: G06T 2207/30181 Earth observation
• CPC Class Codes: G06T 2207/30248 Vehicle exterior or interior
• CPC Class Codes: G06T 7/246 using feature-based methods, e.g. the tracking of corners or segments

Abstract

A navigation system comprises an image sensor operable to obtain range data for a first scene, and a processing unit coupled to the image sensor. The processing unit is operable to identify one or more plane features, based on the range data, using each of a plurality of scales. The processing unit is further operable to combine each of the one or more plane features with a corresponding plane feature from each of the plurality of scales and to project the one or more combined plane features to a reference orientation.

Claims

1. A navigation system, comprising:

an image sensor operable to obtain range data for a first scene; and

a processing unit coupled to the image sensor and operable to identify one or more plane features, based on the range data, using each of a plurality of scales;

wherein the processing unit is further operable to combine each of the one or more plane features with a corresponding plane feature from each of the plurality of scales and to project the one or more combined plane features to a reference orientation.

2. The system of claim 1, wherein the processing unit is further operable to estimate motion by comparing a location of the one or more projected plane features in the first scene with a location of one or more corresponding projected plane features in a second scene.

3. The system of claim 2, further comprising:

at least one alternate navigation sensor operable to provide a motion estimate to the processing unit;

wherein the processing unit is further operable to combine the estimated motion based on the location of the one or more projected plane features in the first and second scenes with the motion estimate provided by the at least one alternate navigation sensor.

4. The system of claim 1, wherein the processing unit is operable to identify one or more plane features by:

calculating a normal vector at each of a plurality of points; and

comparing the difference between the normal vectors at each of the plurality of points with a threshold;

wherein each point, for which the difference is below the threshold, is determined to be a part of a plane feature.

5. The system of claim 4, wherein the processing unit is operable to calculate the normal vector at each of the plurality of points by:

calculating a plurality of cross-products, each between two difference vectors, wherein each difference vector represents the difference between a selected one of the plurality of points and a boundary point of the selected point'"'"'s neighborhood; and

approximating the normal vector at the selected point based on the plurality of cross-products.

6. The system of claim 4, wherein the processing unit is operable to calculate the normal vector at each of the plurality of points by:

calculating a gradient between the center points of adjacent neighborhoods; and

deriving the normal vector from the calculated gradients.

7. The system of claim 1, wherein the processing unit is operable to identify one or more plane features by:

calculating, for a selected one of a plurality of points, a plurality of cross-products, each cross-product between two difference vectors, wherein each difference vector represents the difference between the selected point and a boundary point of the selected point'"'"'s neighborhood; and

comparing the difference between the plurality of cross-products for the selected point with a threshold;

wherein if the difference is below the threshold, the selected point is determined to be a part of a plane feature.

8. The system of claim 1, wherein the processing unit is operable to project the one or more combined plane features to the reference orientation based on the normal vector of the corresponding one or more plane features from the largest scale.

9. A method of performing navigation tasks, the method comprising:

obtaining range data of a first scene with an image sensor;

providing the range data from the image sensor to a processing unit;

identifying one or more plane features in the scene based on the range data;

segmenting the scene based on the one or more identified plane features;

projecting the one or more segmented plane features to a reference orientation; and

analyzing the one or more projected plane features.

10. The method of claim 9 wherein analyzing the one or more projected plane features comprises matching the one or more projected plane features to corresponding projected plane features in a second scene.

11. The method of claim 9, wherein identifying one or more plane features based on the range data comprises:

for each of a plurality of scales, calculating a normal vector at each of a plurality of points in the scene;

for each of the plurality of scales, comparing the difference between the normal vectors at the plurality of points with a threshold, wherein each point, for which the difference is below a threshold, is determined to be a part of one of the one or more plane features;

combining the one or more plane features with corresponding plane features from each of the plurality of scales; and

detecting a local maxima in each of the one or more combined plane features.

12. The method of claim 11, wherein calculating a normal vector at each of the plurality of points comprises:

calculating a plurality of cross-products, each between two difference vectors, wherein each difference vector represents the difference between a selected one of the plurality of points and a boundary point of the selected point'"'"'s neighborhood; and

approximating the normal vector at the selected point based on the plurality of cross-products.

13. The method of claim 11, wherein calculating a normal vector at each of the plurality of points comprises:

calculating a gradient between the center points of adjacent neighborhoods; and

deriving the normal vector from the calculated gradients.

14. The method of claim 11, wherein projecting the one or more segmented plane features to a reference orientation comprises projecting the one or more segmented plane features to the reference orientation based on the normal vector of the corresponding one or more plane features from the largest scale.

15. A program product comprising program instructions embodied on a processor-readable medium for execution by a programmable processor, wherein the program instructions are operable to cause the programmable processor to:

calculate, for each of a plurality of scales, a normal vector at each of a plurality of points in a first scene based on range data received from an image sensor;

compare, for each of the plurality of scales, the difference between the normal vectors at the plurality of points with a threshold, wherein each point, for which the difference is below a threshold, is determined to be a part of a plane feature;

combine the plane feature with a corresponding plane feature from each of the plurality of scales; and

detect a local maxima in the combined plane feature.

16. The program product of claim 15, wherein the program instructions are further operable to cause the programmable processor to calculate the normal vector at a selected one of the plurality of points by calculating a plurality of cross-products, each between two difference vectors, wherein each difference vector represents the difference between the selected point a boundary point of the selected point'"'"'s neighborhood;

and by approximating the normal vector at the selected point based on the plurality of cross-products.

17. The program product of claim 15, wherein the program instructions are further operable to cause the programmable processor to calculate the normal vector at each of the plurality of points by calculating a gradient between the center points of adjacent neighborhoods and deriving the normal vector from the calculated gradients.

18. The program product of claim 15, wherein the program instructions are further operable to cause the programmable processor to:

segment the scene based on the combined plane feature;

project the segmented plane feature to a reference orientation based on the normal vector of the corresponding plane feature from the largest scale; and

match the projected plane feature to a corresponding projected plane feature in a second scene.

19. The program product of claim 15, wherein the program instructions are further operable to cause the programmable processor to convert the range data received from the image sensor to a Cartesian coordinate system.

20. The program product of claim 15, wherein the program instructions are further operable to cause the programmable processor to:

calculate, for a selected one of a plurality of points, a plurality of cross-products, each cross-product between two difference vectors, wherein each difference vector represents the difference between the selected point and a boundary point of the selected point'"'"'s neighborhood; and

compare the difference between the plurality of cross-products for the selected point with a threshold;

wherein if the difference is below the threshold, the selected point is determined to be a part of a plane feature.