Correlative STORM and SEM registration

[jni]

Hi all,

I’m trying to register some STORM and SEM images, samples below:
https://www.dropbox.com/s/04a4gedax8l45qm/CS2_loc2_001_405_01_SRR.tif?dl=1
https://www.dropbox.com/s/rhjs3i4qbktqpbx/cell2_01.tif?dl=1

The main issue is that the STORM resolution is so good that (it appears)
the label (immuno stain) is offset from the actual protein being labeled
(the knobs in the SEM image). Often, you can see that the knob is actually
*surrounded* by the label.

I’ve got an imperfect manual registration, matrix:
[0.37082889676094055, -0.37944599986076355, 0.37944599986076355,
0.37082889676094055, 721.8716430664062, 1532.6571044921875]

and a corresponding (inverse) transformed STORM image:
https://www.dropbox.com/s/778hkrb8rrd6btu/stormt.tif?dl=1

But I haven’t been able to come up with a reliable way of registering the
two images automatically. Most papers I have found use fiducials, which we
don’t have currently. Does anyone have any ideas?

Thanks,

Juan.

--
ImageJ mailing list: http://imagej.nih.gov/ij/list.html