crack detection

OpenMV Products
1

Can OpenMV products detect a line with micrometer thick?
I want to detect a stack (double) wafer and these wafers are very thin 675 micometer by using OpenMV products. What camera will be the best suite for this application?

Thank you

2

It really depends on the optics you use. You’re going to need to make a custom lens M12 mount that can focus at a very close object.

3

Thank you for the response. For start, I will buy OpenMV Cam H7 and M12 lens. Is M12 lens fit on H7?

Thanks

4

Yes, it’s an M12 lens mount.

5

I’ve seen this before, I can ask where to get one if that zoom is enough for you ?

image_from_ios_1024.jpg
image_from_ios_1024.jpg450×600 41.9 KB

image_from_ios_720.jpg
image_from_ios_720.jpg450×600 39.3 KB

2.png
2.png766×600 129 KB

6

00005

Can you give me the specification of the len? By the way, I have a wafer with a notch and cannot liable to detect it. I used keypoints to detect the notch but not consistent. Can you help me or direct me to a direction to fin the notch on the wafer.

Thanks

7

You’re probably going to need more resolution in that image.

The lens info is on the product page.

You can use find circles to find the circle location… then you may wish to walk the circle edge and look for pixel diffs. Alternatively, binary masks methods. The crack is very small… so, to detect it you will need a complex method.

8

Thank you for the response. Do you have any recommendation lens for this purpose? Is there an example how to walk the circle edge? I can use find_edges method to get the circle but don’t know how to walk on the edge.

Thank you

Chi

9

Once you have the circle center and radius you can use the equation for the circle to rotate an angle from 0-360 and use get_pixel() to check each pixel along the edge. You may have to check a region of pixels. You can then sum the region of pixels and check if they are above a threshold. A crack would be under it.

10

Thank you for the response. I use img.find_circles and the center kept moves. Also, I try to use img.find_circles in VGA and it error popup “MemoryError: Out of fast Frame Buffer Stack Memory! Please reduce the resolution of the image you are running this algorithm on to bypass this issue!”

any ideas how to get Find_circle work in VGA mode?

11

Use the OpenMV Cam H7 Plus. The regular has limited RAM.

12

I had this H7 cam worked before but had a higher priority project so i put it on the side. I came back to the project and updated my IDE and now the IDE doesn’t find the camera. I erased the flash and download the firmware as instructed but still not recognizing camera. Please help.

Chi

13

Hi, the camera is fine. Do this:

Put a wire between BOOT0 and RST or VCC.

Plug the camera into the computer USB. Follow the prompts from the IDE. It will flash the bootloader, then tell you to remove the camera from the OC and the wire and then plug it back in to finish loading the firmware.

After you do this you should be good to go. The IDE will walk you through what to do.