03-16-2021 07:56 AM
I want to select every singel spot from an image and I can select most of them, however, still have some spots missing, and I would like to know:
Can I use certain kind of interpolation or algorithm to get the missing spots, its x, y position array?
what I have: selected spots' positions, i.e., 2D array of (x,y).
Thank you.
Solved! Go to Solution.
03-17-2021 04:11 AM
You'd have to explain yourself better...
Explain it like we don't know anything about your application. Because we don't.
What is a missing spot? The black lines? Or a missing green thingy?
Are those green marks placed by you? Should they be placed on the red circles that don't have one? It's just not clear to me.
@Jinghang wrote:Can I use certain kind of interpolation or algorithm to get the missing spots, its x, y position array?
You can probably use a certain kind of interpolation.
I don't see how you can use interpolation to get the missing spots. You could use interpolation to fill the missing spots.
@Jinghang wrote:what I have: selected spots' positions, i.e., 2D array of (x,y).
You have the selected spots?
I understood from the previous line that you wanted to know if you could use interpolation to get the missing spots?
What software are you using? NI Vision? Or just pure LabVIEW?
03-17-2021 10:43 AM - edited 03-17-2021 10:45 AM
Thanks. I will explain it more here.
What do I have?
I have an image with many spots;
I have Vision Assistant and LabVIEW
What do I want?
I want each spot's position, i.e., x,y value in pixel.
What problem do I have?
I could get the position of most spots by using Shape Detection or Particle Analysis (which are marked as Green), but still some ones' x y position are still missing (unseleted by the program, which are red spots with no Green).
What can I do?
what should I do to automatically know the pixel x,y position of the missing (unselected) spots?
03-17-2021 11:05 AM
@Jinghang wrote:What can I do?
what should I do to automatically know the pixel x,y position of the missing (unselected) spots?
Either tune the vision algorithm to catch them all, or find some logic in the grid, and use math\statistics\logic to fill in the gabs. A 3rd option would be to find 2 or more 'imperfect' algorithms that detect spots, and combine the results.
You'd need prior knowledge to come up with algorithm to fill in the missing spots. Are these spots on a grid? A rigid grid? Is it always (more or less) using the same spacing? Can it be rotated? And so on.
You're actually 'faking' measurement results. You'd have to establish the rules for the fake to be acceptable.
03-18-2021 01:16 PM
Thank you for your reply. Your proposed some methods which are very inspiring. I will accept it as the solution. Best regards