Hello Team,
I am writing to inquire about the technical capabilities of the Vision ROI settings and the Pixel Counter for our current project.
We are currently testing equipment designed to inspect foreign substances on the surface of copper plates. During the testing phase, we observed that the plates are not mounted in a fixed position when placed on the conveyor belt. their alignment and position vary slightly each time. The issue we are facing is that when using a fixed rectangular ROI, any tilt or misalignment of the plate causes the system to incorrectly identify those areas as errors, as shown in the attached photo.
To ensure accurate inspection, we need the ROI to be flexible rather than fixed. Specifically, we would like to configure the ROI settings so that they automatically adjust and follow the specific shape and contour of the copper plates. Is there a viable method or feature to achieve this dynamic ROI adjustment based on the object’s shape?
Or Could we used region of blob to pixel counter ROI?
There is not a simple way to make the ROI fit the plate when you have tilt and skew, as far as I know. If it only had been rotation then alignment would have been your friend, but not in this case. The only solution I can think of requires 2 pictures and a bunch of math.
First Image, here you by using probing (two or more for robustnes) determines the plate position using the min/max x/y resultats of the BLOBs
Then you though viAccess changes the ROI to fit. This can be done with the viAccessSetVfExecutionRoi either the Region (each px in the ROI) or the REGION_RLE (each “line” in ROI, in your example this would be the best) and the I would use a slight offset from the calculated edge to minimize the change of including the edge. And you also need take the motion between pictures into account when calculating the ROI position but that should simply be an offset in x or y direction. and you know the speed (I assume) and the image nettime so it simple to calculate.
]