Ask HN: Camera array for small-volume Real-Time passive surveillance of biology | Hacker News Reader

Ask HN: Camera array for small-volume Real-Time passive surveillance of biology

5 pointsJabavuAdams6 comments

If you want to look at C. elegans (1mm long), you use a microscope. But of course, the field of view is small, and you can't just passively surveil a volume of say a few cm^3 and wait to see interesting and possibly unexpected behaviours.

Is there anything like this now? I'm thinking of something similar to wide-area airborne passive surveillance and recording, but for a small area.

What would we learn if we could record a room for a night and see every bedbug? Or record a patch of ground and see every ant? Or record a volume or area of a few cm^3/cm^2 and see everything down to 1 um resolution? Are there currently microscope arrays?

View on HN

Many scopes support region of interest detection. They will use a low-power objective to scan a large field of view, then switch objectives and only look at regions identified from the larger field of view. This really only works for things on slides.

You can definitely couple a macro lens to a DSLR and mount it on a turrent and scan, trading off field of view with time (depends on how fast you can scan, etc).

There are camera arrays (http://graphics.stanford.edu/projects/array/) and an experienced practictioner could take these ideas and do all sorts of amazing imaging.

Great, thanks! Was thinking about COTS image sensors but then the problem becomes how to get the light to them.

Maybe an array of jevois with NoIR lenses and an IR illuminator (http://jevois.org/doc/Lenses.html) with the jevois doing object detection in array mode over a large FoV, then another camera with a much nicer macro lens and high quality sensor (blackfly S w/ a DSLR macro) on a turrent that points to where you see objects of interest identified by the jevois object detector.

Would be a fun project.

Look up Fourier Ptychography, i.e. like this presentation by Guoan Zheng [0]. Fourier Ptychography is basically a variation of Gerchberg-Saxton phase retrieval in the sense that both methods and their variations reconstruct images from constraints. In Ptychography for microscopy they come from low-resolution, small-Field-of-View image captures. You can do superresolution with it and extend your field of view.

Usually in microscopy, you have a very limited field of view but with the variations of this method, people have reconstructed multiple square-mm fields of view with a µm resolution. If you have multiple cameras, you can speed up the process [1]. This also has been done for volumetric samples with the ability to refocus.

You'd probably need high speed cameras and coherent illumination (for best superresolution results).

[0] https://www.youtube.com/watch?v=hece_x37ITg [1] http://computationalimaging.rice.edu/databases/towardcca/

Well, this looks like exactly what I was thinking of...

http://horstmeyer.pratt.duke.edu/gigapixel-micro-camera-arra...

Looks like those Duke people have commercialized this as Ramona Optics.

Ramona Optics https://www.ramonaoptics.com/company

SBIR for High Resolution Parallelized Imaging of Freely Swimming Zebrafish with a Gigapixel Microscope https://www.sbir.gov/sbirsearch/detail/1323251

Ramona Optics SBIR info https://www.sbir.gov/node/1155319

Others: https://www.semanticscholar.org/paper/Wide-field-microscopy-...

https://www.osa.org/en-us/about_osa/newsroom/news_releases/2...

https://www.osapublishing.org/oe/fulltext.cfm?uri=oe-21-2-23...

Reply on news.ycombinator.com