US 9,811,880 B2 | ||
Backfilling points in a point cloud | ||
Terrell Nathan Mundhenk, Calabasas, CA (US); Yuri Owechko, Newbury Park, CA (US); and Kyungnam Kim, Oak Park, CA (US) | ||
Assigned to THE BOEING COMPANY, Chicago, IL (US) | ||
Filed by The Boeing Company, Chicago, IL (US) | ||
Filed on Nov. 9, 2012, as Appl. No. 13/673,429. | ||
Prior Publication US 2014/0132733 A1, May 15, 2014 | ||
Int. Cl. H04N 7/18 (2006.01); G06T 3/40 (2006.01); H04N 13/00 (2006.01); G06T 5/00 (2006.01); H04N 13/02 (2006.01); G06T 17/00 (2006.01) |
CPC G06T 3/4007 (2013.01) [G06T 3/4038 (2013.01); G06T 5/002 (2013.01); G06T 17/00 (2013.01); H04N 13/0022 (2013.01); H04N 13/025 (2013.01); G06T 2210/56 (2013.01)] | 17 Claims |
1. An apparatus for improved resolution and accuracy of an object in a scene in a point cloud, comprising:
an image processing system comprising a processor configured to:
receive a two-dimensional image of a scene from a camera system and a point cloud of the scene having a first resolution from
a light detection and ranging system;
create transformed points by mapping point locations for a portion of points in the point cloud describing a three-dimensional
representation of the scene to corresponding pixel locations in the two-dimensional image of the scene to form the transformed
points;
create a fused data array using the two-dimensional image and the transformed points, wherein each transformed point corresponds
to a pixel in the two-dimensional image and to a matched element in a number of matched elements in the fused data array;
calculate a goodness score using a combination of a distance of a transformed point mapped to a pixel corresponding to the
matched element from the camera system, a dissimilarity of the matched element to other matched elements wherein the goodness
score increases as the dissimilarity increases, and a depth value in a data vector associated with the matched element, wherein
the goodness score is generated based on a relationship where:
![]() and
Δi=di o·γ,
wherein:
G i is the goodness score of an ith matched element;
Mi is a dissimilarity score for the ith matched element;
Δi is a distance measurement for the ith matched element;
j is an index for an n matched elements;
F is a response to a feature;
m is a number of features;
dio is the depth value in a data vector associated with the ith matched element; and
γ is a normalizing constant;
use the goodness score to identify a number of support elements; use the fused data array and the number of support elements
to identify new points for the point cloud;
add the new points to the point cloud to form a new point cloud, wherein the new point cloud has a second resolution greater
than the first resolution; and
identify the object in the scene using the new point cloud.
|