I’m planning on doing a project that involves pulling a 0.3MP JPG image over USB from an external device and then doing some basic image processing in python before spitting out some summary statistics on a LED screen on the pi.
I was hoping to get some input on how much memory I would likely need on my Pi for a project like this. It’s only processing one image at a time so Im hoping I can get away with a Model B 2Gb rasberry Pi, but I was wondering y’all think it’s worth shelling out for more memory on board? (Or if I’m better off with a completely different bit of kit haha)
I’ve done some basic (but slow) image processing on a Pi Zero, so I’d say a 2GB would be fine. I’d look into OpenCV as it’s trivial to use with Python and supports a ton of functions.
Depending on the complexity and speed required for your image processing, you might need to step up to a Jetson Nano, and find some GPU accelerated libraries that’ll run on the Maxwell GPU.
I’d love to hear about what you’re doing with this to get a better idea of what parts to recommend, so if you feel like writing up your application, that’d be great!
Hey James. Thanks for your advice! I haven’t written any of the software yet, I’m just getting started. But the general idea is to measure the size and variance in particle size of some particulate matter.
So I am expecting to use Sobel edge detection algorithm (or similar) for the software to identify particles and then calculate the area of the smallest circle that encloses that particle as an estimate for that particle volume. Repeat for maybe a few dozen particles.
It may even be easier for me to conbert the image to a binary black/white map to identify the particles.
If that helps you give any more specific advice.
Sounds super interesting! What’s the range of sizes that you need to detect with your camera? There are sensors specifically made for detecting particles, but I’m imagining that you’re making up this project for the sake of interest or education rather than ease.
Keen to see this happen!