Dear Thiago,
I would suggest you to check the alignment at each step. Use applyxfm
for the scan1_to_scan2 and scan2_to_scan1 to see if your two images
are aligned together in both of these registrations. Also, I don't use
the -v option in the midtrans call so I don't know if it affects the
results somehow. The last thing that you could check is, if the XX0002
matrix's non-diagonal elements are half the scan1_to_scan2 elements
(XX0001 half of the scan2_to_scan1 as well). This might help you to
spot what causes this inconsistency.
Hope that helps,
Vasilis
On 11 July 2014 21:07, Thiago Junqueira Ribeiro de Rezende
<[log in to unmask]> wrote:
> Dear Vasilis,
>
> Thanks by the suggestion. However, I couldn't solve the problem. The resulting images still have inconsistent orientations.
> Anyone has any other suggestion?
>
> All the best,
>
> Thiago
|