Raspberry pi cameras for mapping?

It’s been interesting to see the improvement in small camera modules available for the Raspberry Pi. I was wondering if anyone had any experience with these for drone mapping? Especially any comments on the impacts of a rolling shutter when using a fixed-wing platform, and any issues with small sensor size and focal length resulting in too small a ground footprint for each image.


What is your use-case?

12.3MP should mean about 4.x cm/px at 400FT AGL, which was fine for my agricultural field mapping, and that sensor is about the same dimension as what is in my GoPro. I didn’t have terribly much trouble with it.

That 64MP one is… eh. That’s going to be a challenge, I think. The pixel-binning sensors really have not impressed me much as they don’t seem to actually resolve any more real information than the base resolution (16MP in this case [quad-bayer]).

The 2MP one could be interesting for really fast data acquisition without distortion, but I’m not sure how usable 2MP would be when reconstructing.

1 Like

Use case is probably just mapping for OpenStreetMap type tracing. Possibly for derived products for more advanced community disaster risk reduction exercises (e.g. slope for landslide risk, DEM for flood modeling).

Good to know about the pixel-binning. The 40% off that they’re running right now had caught my attention.

1 Like

There is a real use-case for the pixel-binning sensors if they operate like Fuji’s older EXR sensors which pixel-binned, in that the sensors could optinally do sensor-domain multi-exposure (at a lower resolution) for HDR/Dynamic range, which allowed those tiny little sensors to really capture some crazy range without clipping.

I think in a mapping context that would be much more valuable than inflated MP counts, but MP counts are ultimately what sell.

1 Like

I agree with Saijin_Naib on most of this, especially inflated MPs not really delivering. Physics eventually catches up in these form factors.

For general OSM tracing, the 12 or 16MP Hi Quality setups they have are fine. If you want elevation models, you want a global shutter and fixed lens. Otherwise: sadness.

ArduCam have M12, C, and CS-Mount for a lot of different backs, and I have been watching them for a while. I have a design for a flyable array of their C/CS mount (or RPi foundation’s equivalent) that I sent to TZ over a year ago. I don’t know if it ever got airborne though, and that rolling shutter would have been a challenge if we were doing a fixed wing (planned for quad flight).

But, these are small enough, I think you could easily do dual mount: one 12-16MP rolling shutter for orthophotos and one global lens for elevation models on the same drone. The target resolution for hydrologic modeling is 5m, though your data collect should be higher for the sake of adequate ground views and filtering. You could even mount 4 of the 50-degree version to maximize resolution and view angles and synchronize using GitHub - localdevices/libre360 or try out their hardware synchronized versions (though those are limited to 1MP).

edit: I think the 2MP’s should work fine. We handle thermal sensors now, so 2MP isn’t a problem, as long as it meets your spec for elevation model resolution (1/2 to 1/4 linear resolution of your orthophoto resolution).