Marius PopescuThings I've Built
Mechanically Microscanned Thermal Imager (MMTI)
Stepper Motor actuated device that pans / tilts a MLX90620 (16x4) IR FPA in sub-pixel shifts, takes 121 low-res frames (16x4) and combines them in a high-res (176x44) thermal image. All processing done live on an iOS device con. via WiFi.
Projects I Like & Follow
Neil Jansen
fl@C@
i.abdalkader
peter jansen
ThunderSqueak
Benchoff
Lutetium
Radu Motisan
RoGeorge
Alan Kilian
Paul Stoffregen
Bruce Land
Aleksandar Bradic
Lyon
Mike Szczys
Radu ConstantinShare this profile
ShareDiscussions
As to the HTPA, I'll set one aside for you! However, we are not likely to get funded this time around. If we don't make it, I'm not sure what our next strategy will be (I've narrowed down to three options - but one of those is give up and I've ruled that out) and won't decide until next week, a few days after we end. If we make it, then we go to order. I'm putting some new videos up tomorrow and hopefully some press and the new videos will get us there.
Great job on hacking a massive improvement to the hi-jack data rate. Other people will also be interested in these other aspects of your project, especially given how robust the protocol already is at this time. I enjoyed your video, wish we'd met two years ago!
If you're interested in a bare HTPA array, let me know.
Erik
Shameless plug: actually except from the great overall concept, I'm not using anything more from the Hijack project at this point :p
Are you even asking? I'd love to evaluate the performance of the HTPA arrays! :)
Well, we can catch up for the two missed years then :))
Let me know how I can be of help w/ anything Hema-Imager related or otherwise.
The great coincidence is that I was planing to contact you about some opinions on the HTPA series anyway.
Great tips on getting microscanning working. I tried $3 hobby servos before discovering your project and am learning a lot about how they don't work for this application :)
Another question I have is the algorithm for deconvolution. I had some success with Richardson Lucy algorithm (http://scikit-image.org/docs/dev/auto_examples/filters/plot_deconvolution.html), but it has a few pixels of ringing around the edges. Are you using a different algorithm or doing border handling? https://en.wikipedia.org/wiki/Kernel_(image_processing)#Edge_Handling
It'll be good for me to work on the above questions myself, but it's nice to learn some tricks too. Thanks!