Dear colleagues,
thank you for your support, in particular to Stephan Preibisch and Nicolas Chiaruttini .
I integrated the updated source code executable as a project in eclipse, thanks for the link.
The aim is image-mosaicing and for calculation of translation, rotation, scaling and feature-matching SIFT-algorithm is used. To produce a mosaic-image with some small single frames works well provided that the single frames are only displaced.
But if the images are additionally rotated, the result is very bad.
Therefore I have some questions and ask for your support:
The source code calculates a matrix in terms of: {m[0], m[2], 0, m[4]}; {m[1], m[3], 0, m[5]}; {0, 0, 1, 0}; {0, 0, 0, 1}
In my opinion m[4] and m[5] are displacement-vectors.
But is there additionally rotation (single frames could be rotated), these objects (m[4] and m[5]) have very great values. The single frames are a long way away from each other and dont touch or overlap in the mosaic.
How do I have to understand the matrix? Do I have to execute a inverse-matrix in dependence on the calculated values of the matrix? Are the dimensions "pixel" or has to be considered a scaling factor? Is the rotation corrected automatically by the algortihm or have I to do that subsequently before stitching? Could someone understand and explain the great values of translation? Do I have to divide them with a factor?
Thanks in advance for your support,
best regards,
Michael
--
ImageJ mailing list:
http://imagej.nih.gov/ij/list.html
Locked
