a projective transform, stored as a homography
What a tilted specimen does to a map: the plane is seen obliquely, so straight lines stay straight but parallel ones need not, and the scale varies across the frame. Eight parameters, fitted by the direct linear transform and solved robustly.
Unlike a polynomial this is a genuine geometric map, so its inverse is another homography rather than an iteration.
Syntax
T = spatialTransformProjective(H)
T = spatialTransformProjective.fit(posA,posB)
T = spatialTransformProjective.fit(posA,posB,'weights',w)Input
| H | 3x3 homography, scaled so H(3,3) = 1 |
| posA, posB | vector3d, the same points in the two frames |
| w | double, one weight per point |
Output
| T | spatialTransformProjective |
Class Properties
| H | 3x3 homography acting on [x; y; 1], divided through by the third row |