Stereo images from Juelich

We are running in colaboration with the Meteorological institute at the university Bonn a pair of cloud cameras with the intention for a 3-D reconstruction of the cloud field (see Beekmans et al. 2016). Here you see the very last images.

Left: Camera at JOYCE, right MIUB camera several hundred meters away.
The JOYCE camera is aligned to north and images are mirrored such that north is on top and east on the right.
(Note that this is the common coordinate system if you are looking on a map but it is not what the camera sees).
The MIUB camera is not aligned, north is somewhere in the lower left corner and east somewhere in the lower right corner.

crossed-eye-view ?

This is a try for a crossed eye-view:
Right image (MIUB-camera) first flipped (top ↔ down) and rotated by 37 degrees (=> north on top and east on the right).
Then both images are rotated by 76 degrees such that rows in the images are parallel to the connecting line between both cameras.
I derived angles empirical - so it might be not perfect. For an explanation of crossed-eye-view see here.

stereo analysis

Left: Undistorted Image of the Juelich Camera. Not that this image is flopped such that east is on the left and north on top. Right: derived height map based on analysis of the two images in the same coordinate system with east on the left as the picture (see Beekmans et al. 2016). You may want to compare with the current data from ceilometers Vaisala ct25k and Lufft CHM15k.

Recitified images of the two cameras. They are the base for the stereo retreival. Try crossed eye view to see 3D.

animated composite

Move the mouse over the image ...

Stripe: every line represent the center scanline of an composite from above.
In principle you sould be able to use crossed-eye-view to see 3D features.