You can check out automatic stitching tools for automated alignment. Photomerge tool in photoshop yields layered aligned inputs, which should meet your needs.
Your figure seems to be made with matlab, if you wish to code entirely in matlab, matlab also has a auto panoramic module.
Your figure seems to be made with matlab, if you wish to code entirely in matlab, matlab also has a auto panoramic module.
I think his goal is to automatically adjust color of scans to match the digital/sample?kiowa said:
You can check out automatic stitching tools for automated alignment. Photomerge tool in photoshop yields layered aligned inputs, which should meet your needs.
Your figure seems to be made with matlab, if you wish to code entirely in matlab, matlab also has a auto panoramic module.
Aligned inputs would be required to create an accurate look-up table, if I understand this right.fireattack said:
I think his goal is to automatically adjust color of scans to match the digital/sample?
Oh yeah you're right, didn't read the second last paragraph carefully. You can actually use the "align the layers" tool under "Edit" menu directly, it's based on the same technology of photomerge though.kiowa said:
Aligned inputs would be required to create an accurate look-up table, if I understand this right.
eccdbb
color lookup tool in Photoshop CS6/CC
https://www.youtube.com/watch?v=VQx4wLr7yCM
https://helpx.adobe.com/speedgrade/how-to/speedgrade-shot-matching-cc.html
I am thinking of making a tool that takes does something similar. Given a input set of (r,g,b) points:
f1(r,g,b,tau) + Ir = r2,
f2(r,g,b,tau) + Ig = g2,
f3(r,g,b,tau) + Ib = b2,
where Ir, Ig, Ib are identity functions for r,g,b
and f1, f2, f3 are adjustment functions " " using weighted sum of radial basis functions (gaussian, currently). See:
http://www.mediafire.com/view/4j9djvd2ez7ib3d/RGB_color_deform%202D.png
I'll probably update in a week after exams...
The only major issue I have now is automating alignment of the reference image w/ scanned image and creating a widely distributed set of input points, but maybe this can be bypassed.
http://www.mediafire.com/view/v69c7auh1xd4fxe/scan%20vs%20sample%20ref.png
This one was impossible to match using curves.