Image processing for pose estimation
A method for estimating a pose of a first object in relation to a second object, the second object comprising a visual marker comprising a plurality of ellipses comprises capturing a video image of the visual marker with an image capture device on the first object, then pre-processing frames of the video image on a graphics processing unit. The method comprises detecting the visual marker by finding contours in the frames to identify the plurality of ellipses and determining that a pattern of the plurality of ellipses match a known pattern of the visual marker. Then the method comprises obtaining coordinates of two or more of the plurality of ellipses of the visual marker, estimating the pose of the first object in relation to the second object by inputting the coordinates of the plurality of ellipses into a pose estimation algorithm, and filtering results of the pose estimation algorithm.
1. A method for estimating a pose of a first object in relation to a second object, the second object comprising a visual marker comprising a plurality of ellipses, the method comprising:
capturing a video image of the visual marker with an image capture device on the first object;
pre-processing frames of the video image on a graphics processing unit, detecting the visual marker at a first distance by:
finding contours in the frames to identify the plurality of ellipses; and
determining that a pattern of the plurality of ellipses match an existing, known pattern of the visual marker;
obtaining coordinates of two or more of the plurality of ellipses of the visual marker;
estimating the pose of the first object in relation to the second object by inputting the coordinates of the plurality of ellipses into a pose estimation algorithm;
filtering results of the pose estimation algorithm;
wherein the visual marker comprises a smaller replica visual marker of itself at a center portion of the visual marker, and wherein the method further comprises:
detecting the smaller replica visual marker at a second distance, the second distance being shorter than the first distance; and
estimating the pose of the first object in relation to the smaller replica visual marker.
2. The method of claim 1 , wherein the first object is a unmanned aerial vehicle (UAV).
3. The method of claim 2 , wherein the image capture device and the graphics processing unit are mounted on the UAV.
4. The method of claim 3 , wherein the results of the pose estimation algorithm are used to control landing of the UAV.
5. The method of claim 1 , wherein determining that a pattern of the plurality of ellipses match a known pattern of the visual marker takes place on the graphics processing unit.
6. The method of claim 1 , wherein finding the contours in the frames to identify the plurality of ellipses takes place on a central processing unit associated with the graphics processing unit.
7. The method of claim 1 , wherein the visual marker comprises one outer ellipse and at least four non-concentric inner ellipses inside the outer ellipse.
8. The method of claim 7 , wherein the visual marker and the smaller replica visual marker each comprise at least four non-concentric inner ellipses inside an outer ellipse.
9. A system for estimating a pose of a first object in relation to a second object having a visual marker comprising a plurality of ellipses, the system comprising:
a central processing unit;
a graphics processing unit; and
an image capture device, the system being configured to:
capture a video image of the visual marker with an image capture device;
pre-process frames of the video image on a graphics processing unit, detect the visual marker at a first distance by:
finding contours in the frames to identify the plurality of ellipses; and
determining that a pattern of the plurality of ellipses match an existing, known pattern of the visual marker;
obtain coordinates of two or more of the plurality of ellipses of the visual marker;
estimate the pose of the first object in relation to the second object by inputting the coordinates of the plurality of ellipses into a pose estimation algorithm;
filtering results of the pose estimation algorithm;
wherein the visual marker comprises a smaller replica visual marker of itself at a center portion of the visual marker, and wherein the method further comprises:
detecting the smaller replica visual marker at a second distance, the second distance being shorter than the first distance; and
estimating the pose of the first object in relation to the smaller replica visual marker.
10. The system of claim 9 , wherein the first object is a unmanned aerial vehicle (UAV).
11. The system of claim 10 , wherein the image capture device and the graphics processing unit are mounted on the UAV.
12. The system of claim 11 , wherein the results of the pose estimation algorithm are used to control landing of the UAV.
13. The system of claim 9 , wherein determining that a pattern of the plurality of ellipses match a known pattern of the visual marker takes place on the graphics processing unit.
14. The system of claim 9 , wherein finding the contours in the frames to identify the plurality of ellipses takes place on the central processing unit.
15. The system of claim 9 , wherein the visual marker comprises one outer ellipse and at least four non-concentric inner ellipses inside the outer ellipse.
16. The system of claim 15 , wherein the visual marker and the smaller replica visual marker each comprise at least four non-concentric inner ellipses inside an outer ellipse.
17. A non-transitory, tangible computer readable storage medium, encoded with processor readable instructions to perform a method for estimating a pose of a first object in relation to a second object having a visual marker comprising a plurality of ellipses, the method comprising:
capturing a video image of the visual marker with an image capture device;
pre-processing frames of the video image on a graphics processing unit, detecting the visual marker at a first distance by:
finding contours in the frames to identify the plurality of ellipses; and
determining that a pattern of the plurality of ellipses match an existing, known pattern of the visual marker;
obtaining coordinates of two or more of the plurality of ellipses of the visual marker;
estimating the pose of the first object in relation to the second object by inputting the coordinates of the plurality of ellipses into a pose estimation algorithm;
filtering results of the pose estimation algorithm;
wherein the visual marker comprises a smaller replica visual marker of itself at a center portion of the visual marker, and wherein the method further comprises:
detecting the smaller replica visual marker at a second distance, the second distance being shorter than the first distance; and
estimating the pose of the first object in relation to the smaller replica visual marker.
18. The non-transitory, tangible computer readable storage medium of claim 17 , wherein the first movable object is a unmanned aerial vehicle (UAV).
19. The non-transitory, tangible computer readable storage medium of claim 18 , wherein the image capture device and the graphics processing unit are mounted on the UAV.
20. The non-transitory, tangible computer readable storage medium of claim 19 , wherein the results of the pose estimation algorithm are used to control landing of the UAV.
21. The non-transitory, tangible computer readable storage medium of claim 17 , wherein determining that a pattern of the plurality of ellipses match a known pattern of the visual marker takes place on the graphics processing unit.
22. The non-transitory, tangible computer readable storage medium of claim 17 , wherein finding the contours in the frames to identify the plurality of ellipses takes place on a central processing unit associated with the graphics processing unit.