Hello.
I am currently doing a calibration project. This involves a robotic arm and a M7 camera. To map the pixels to the robotic arm coordinate system, I tried to get an understanding of the M7’s field of view. This is what I obtained:
The polygon that can be viewed here represents the image boundaries of the M7. I used the single color detection example available on the openMV IDE and moved a colored object along the boundaries.
What I would like to know is:
- Is this the normal behaviour of the camera?
- If not, can it be fixed?
- If it is, is there any mathematical formular to represent this capture?
Was the camera pointed straight down or at an angle? Also, did you use any lens correction to deal with the lens roundness on the edges?
Hello.
Yes, this was done with the camera pointing straight down, it was attached to a level surface (the surface is parallel with the captured surface).
The lens in usage is the IR Lens.
Hmm, I don’t really know what’s going on. Our camera’s aren’t calibrated parts. I’d say these two things are happening however:
- The lens applies some amount of image distortion which you should try to correct for using the lens_corr() method. If just takes a strength number and you should manually inspect the image to see if the lines in it are straight after using the method.
- After doing the above, the lens mount may be slightly miss-aligned. There’s play in the lens mount which allows the image circle to not be projected perfectly on-top of the camera IC.
Other than the above I do not know what’s causing the issue.
