iRVision 2D, How to find part deviations from original taught image

    Hey guys, I am running a 2d inspection process of flat blanks of steel running traveling through a 2d vision camera, which basically line tracks with a robot to pick said parts off of a conveyor and stacks them in place.


    The blanks are well lit and are flat on a black conveyor, very easy for the vision system to see the parts. How ever i recently had a few parts go through the system that had large holes in the blanks of steel which is a deviation of what the part should look like. My taught image is solid and only sees the outside edge of the entire part. I assumed the vision system would see the holes with in the part and reject it but it seems based on my taught part it only cares about basically the outside edge of the part.


    Is there a way to teach my GPM locator tool (or other tool) to look for holes or other deviations and send a fail signal rather than a pass ? it seems i can teach to look for these deviations and add them as requirements for the part to pass but i want to do the opposite. The holes in the parts can also land any where on the part, not just the same spot over and over.


    Appreciate any advice, thanks.

    Do you have an iRVision image of a part with holes that you can share? You can probably use one of the inspection tools to monitor how many light vs dark pixels are in an image if there happens to be a significant difference between the holes and a solid part.

    Jake
    I know exactly what you are talking about :uglyhammer2:


    Are the holes same diameters ?
    What size holes are we talking about here ?
    Do they show completely random or on one particular side or corner of the part ?


    The blob is a good idea, You could measure the number of white or black pixels of the perfect part, find and average because your parts shine a bit and then everything over or under a range of your tolerance will be a bad part


    You can add this tools without destroying what you have. Still do a backup and add the tool

    Retired but still helping

    Thank you for the responses, sorry for the delay i've been sucked into a few other projects :censored: . The holes in this particular part were about an inch wide. They also can potentially appear any where on the part not the same spots. Attached is an image of my recipe for this particular part. All my parts are exactly the same recipe to recipe. They are all some sort of solid shape, flat blank of steel. I did also think i could some how use the blob locator to find these holes but again, not sure how to teach it to be a failed part rather than a requirement for a pass. Whatever tool i choose to use i need to be able to replicate across every part for the line.


    UPDATE :


    The blob locator tool worked. I was able to see the lack of pixels in an image with a hole on a part versus an image with no hole in a part. A half inch NPT knock out created a 100~ pixel difference. The new problem i am encountering is that i still need to use the GPM locator tool for orientation of the part. When a part with no hole goes through the vision system, the robot will pass on the blob and gpm tools and pick the part no problem. when a part with a hole goes through however, it fails on the blob tool but passes still on the gpm tool, and thus the robot will still pick the part, which is not ok.


    Any other thoughts ? is there anyway to make my vision conditional to both tools passing instead of either or ?

Advertising from our partners