search

alicevision / Meshroom

3D Reconstruction Software
http://alicevision.org
Other
11.11k stars 1.08k forks source link

Ability to use SEM images? #436

Closed Gorecraft closed 5 years ago

Gorecraft commented 5 years ago

Hello, i want to use scanning electron microscopy images and therefore i don't have a camera sensor but a scanned area and an electron detector. What informations should i provide for "camera intrinsic"? focal length is more or less the working distance and sensor width the "true" width of the image (width of the scanned area)? Thanks and regards

natowi commented 5 years ago

Did you try processing the images with default values? I was able to reconstruct SEM images from screenshots (!) of images from this paper sem 3D Demo: https://skfb.ly/6JozU

If this is not accurate enough for you,I think you should be able to estimate the "camera intrinsic" parameters. There are a few papers discussing SEM photogrammetry and camera calibration such as: http://www.irisa.fr/lagadic/pdf/2014_icra_cui.pdf, http://www.irisa.fr/lagadic/pdf/2015_ijom_cui.pdf or https://hal.archives-ouvertes.fr/hal-01644885/document

If you have a calibration pattern, you could maybe use it to get these parameters as well. https://github.com/alicevision/meshroom/issues/921 https://boofcv.org/index.php?title=Tutorial_Camera_Calibration https://docs.opencv.org/2.4/modules/calib3d/doc/camera_calibration_and_3d_reconstruction.html It would be interesting to see how a calibration pattern works on this scale.

natowi commented 3 years ago

Microphotogrammetry experiments if someone is interested: http://fab.cba.mit.edu/classes/864.20/people/camron/microphotogrammetry/final.html