Pi pico OV5640 camera calibration with openCV

Print a target.
Hack some python
Solder yet more wires
Take some photos.
Open CV blob detection and then chessboard finding (yes, really)
Calculate and save distortion matrices.
Use saved distortion matrices to un-distort any images taken on this camera.
What does this mean? I'm not quite sure yet. But it feels like progress.
#makerdad #3dprinting #3dscanner #pico #ov5640 #opencvpython

[@StormBurnX]

What a lovely project! If you're curious, the fisheye distortion adjustment makes a lot more sense when you see the transition from no fix to full fix, rather than only seeing the before/after. not sure if I can share links here but the youtube video "J7hdrMn2RII" shows a short animation (about 15 seconds long) of how fisheye correction works.
If your lens is fixed focal length, that is to say, if you can calibrate the camera sensor once and then never need to re-calibrate it, you will find it immensely handy to point the camera at a large, blank wall so that there are no ceilings/corners/floors/etc in the field of view, only the wall itself, and then progressively move the calibration sheet along the wall to calibrate the camera's entire field of view. This will greatly aid in fixing the distortion that's particularly notable around the far edges of the camera's view, and by placing the calibration targets on a wall, they will remain co-planar for the entire calibration process, which greatly increases the quality of the software's output.
If it helps, you can even sort of divide the wall into rows/columns like a spreadsheet, to ensure you capture calibration images uniformly along the entire wall, and thus entire field of view :)
As an alternative, depending on which setup you're using for blob detection and chessboard search, you might actually find it even easier and faster to do the aforementioned setup, but print 5 target sets, putting one in each corner of the camera's view and the fifth as close to center as you can reasonably manage, and then run the distortion matrix calculator off that single image with all 5 co-planar target papers. This works quite well for rapid, uniform calibration, but I'm not sure if that is built into the code/modules you're using, so you may still have to do each position manually, but all the same - doing them along a wall will greatly assist the un-distortion :)

[@fleurysrajoelisoaandrianir3894]

Hello. ReaLly need your help. Actually working on a projet (3D scanner). Using OV5640, a laser line, and RaspBerry Pico. Now having issue with memory lack, when increasing image size. I see that you use a memory card. Can we store variable directly inside of it, without allocating space inside RAM?

[@fleurysrajoelisoaandrianir3894]

Could you capture images with more than 240X240 size? If yes, could you help, or share part of your code.

[@denisseyemirelbautistavivanco]

Hi! thanks por the information at the video, i´d just like to ask if yo could share the programation in other video please