Package boofcv.alg.fiducial.calib.circle

Class EllipseClustersIntoHexagonalGrid

java.lang.Object
boofcv.alg.fiducial.calib.circle.EllipseClustersIntoGrid
boofcv.alg.fiducial.calib.circle.EllipseClustersIntoHexagonalGrid
public class EllipseClustersIntoHexagonalGrid extends EllipseClustersIntoGrid

Given a cluster of ellipses (created with EllipsesIntoClusters) order the ellipses into a hexagonal grid pattern. In a hexagonal grid the center of each circle is the same distance from its neighbors. Smallest grid size it will detect is 3 x 3.

Note that the returned grid is 'sparse'. every other node is skipped implicitly. This is caused by the asymmetry. Each row is offset by one circle/grid element.

Examples:
 3x6 grid will have 9 elements total.
 grid(0,0) = [0]
 grid(0,2) = [1]
 grid(0,4) = [2]
 grid(1,1) = [3]
 grid(1,3) = [4]
 grid(1,5) = [5]

IMPORTANT: To properly construct the contour the clusters need to connect and jump over the "zig-zags". Thus at a minimum search radius should be distance between centers*2 if not more.

See EllipseClustersIntoGrid.Grid for a description of how the output grids are described. It uses a sparse format.

See DetectCircleHexagonalGrid for an example of an hexagonal grid