This forum is disabled, please visit https://forum.opencv.org
 | 1 | initial version |
mapping depends of linTr_ Mat and shift_
| | 2 | No.2 Revision |
mapping depends of linTr_ Mat and shift_shift_
see sample
| | 3 | No.3 Revision |
| | 4 | No.4 Revision |
mapping depends of linTr_ Mat and shift_ It is a constant map because there is no method to change transfomr. You can set affine transform only in constructor (You can compose current map with another map)
see sample
All is in sample :
Simulate a transformation img1 (reference) in img2 new image img2= [linTr shift] img1
Matx<double, 2, 2> linTr(1., 0.1, -0.01, 1.);
Vec<double, 2> shift(1., 1.);
MapAffine mapTest(linTr, shift);
mapTest.warp(img1, img2);
Now we want to find a transformation between img1 and img2. Of course projTr is unknown
MapperGradAffine mapper;
MapperPyramid mappPyr(mapper);
Ptr<Map> mapPtr;
mappPyr.calculate(img1, img2, mapPtr);
we find an estimation of projTr
--- Testing affine mapper ---
real --> [1, 0.1;
-0.01, 1]
[1;
1]
estimated --> [0.9999473655589267, 0.1009308875914515;
-0.009985648108361955, 0.9998113921559071]
[0.9551427088080436;
1.024064674225132]
and finally we transform img2 in dest to check difference between img1 and dest. We must inverse transformation using inverseWarp :
Mat dest;
mapAff->inverseWarp(img2, dest);
showDifference(img1, dest, DIFF_REGPIX_IM);