computer vision - How to match orientation and scale of two different image of the same object in OpenCV? -
i have 2 images of printed circuit boards (pcb) both showing same pcb. differences between them lighting, scale , orientation (because take pcb images phone camera).
now want use 1 image of pcb check if components of circuit assembled on identical pcb.
is there convenient way check differences between 2 images of 2 identical pcb?
btw, can add marks on pcb in opencv can correct orientation , scale of image.
pcb = printed circuit board, right?!?
you compute projective projective transformation or homography between matched points in both images. transformation can used match planes (like pcbs) , considers scale, rotation, shear , projective changes between images.
it's simple method:
select @ least 4 points , solve system of linear equations. take @ answer question on math se explains that.
this opencv example uses (automatic) feature matching find corresponding image points , computes homography.
the interesting derivation of transformation can found in every computer vision text book, e.g. standard zisserman's "multiple view geometry" or ma's "an invitation 3-d vision".
edit:
this method not remove specular reflections or other intensity differences.
Comments
Post a Comment