Printable Version of Topic

Click here to view this topic in its original format

Unmanned Spaceflight.com _ Image Processing Techniques _ Interesting camera design for future microscopic imagers?

Posted by: hendric Jun 22 2011, 07:38 PM

Found this on Facebook:

http://blog.lytro.com/news/the-light-field-revolution-by-ren-ng/

His dissertation is available here

http://www.lytro.com/renng-thesis.pdf

Being a non-expert, I only skimmed it for the pictures, but it looks like that it uses the extra resolution available in today's camera chips to increase the focused depth of field using light rays.

Microlenses are placed in front of a group of pixels, say 8x8 or 64, and now the mini-picture formed on that 8x8 array can be used to extend the Depth of field by a factor of 8. This can be done without needing a high F/ratio, as is usual to increase the DoF. Plus, it allows for choosing to put near, far, or everything in between into focus.

For space, this looks the most interesting for microscopic images. It would allow a single-shot to capture a whole 3d microscopic image, with focus at a very wide distance range. You could create the large DoF picture on the imager, and upload only the processed image, or upload the full image, and choose where to focus back on Earth. It would reduce your effective number of pixels, but would remove the need for focusing apparatus, aperture control, or focus steps as you bring the imager closer to the object under study.

Posted by: jekbradbury Jun 22 2011, 07:59 PM

The conceptual idea has been around for about 20 years, but it's only now that sensors with the required resolution (around 100 megapixels to get a resulting focused image at a reasonably high resolution) have become available. Unfortunately, such a large amount of data (around the size of a full-resolution MER panorama, if I'm not mistaken, for a single image) would make the mission limited quite severely by downlink bandwidth.

Posted by: djellison Jun 22 2011, 08:05 PM

MAHLI will generate an single extended DOF image and a height map, on the back-electronics for the camera, before transmitting those to the ground. It need a focusing mechanism, but it doesn't need a month to download it

Posted by: hendric Jun 22 2011, 10:17 PM

Well, I was thinking the DoF processing could be handled on-site, as an overnight activity. After all, if you're going to need heaters, might as well use the power to run CPU cycles to generate heat.

I guess one possible issue would be 100+ MP chips are going to require pretty small pixels dimensions, and smaller pixels are more likely to suffer with CR hits etc. But to offset, the pictures would be much shorter; the examples in his thesis are F/4, with comparable DoF as F/22. Shorter exposures would limit the number of CR hits, right?