Hi,
if I try your syntax
avwmaths++ -test_img -kernel file kn -fmean test_img_mean
the routine causes an memory access error. If I change the position of
-fmean and -kernel
avwmaths++ -test_img -fmean -kernel file kn test_img_mean
it seems to work on a 32-bit system and takes only a few seconds. On
64-machine both versions produce an error.
Seems to work means: I get an output but if I try the same with some
included kernels like box and gauss with serveral widths, the output image
is allways the same, a little bit smoothed. Its independet from the kernel
parameter.
If I try avwconv it works and the procedure takes a long time, with
--forcefft too. This seems more real to me than a few second.
Best regards
Jörg
|