Advanced edge detection - steepness/slope issue

[kaka21]

Hello all!

I am using Advanced edge detection to find the edge points of a gap between two objects as shown in the images attached. In both the images the distance between the two edge points are almost equal, which doesn't make sense because in one image the white level of the gap is higher than the other image. This indicates that the gap is bigger in the image that has higher edge strength than the image that has a lower edge strength. I tried to manipulate the steepness of the edge detection parameter by increasing the kernel size, but it did not work out well as the gap is just 2 pixels to 3 pixels wide. Is there any method to measure the gap based on the steepness or slope value of the edges for objects that are less than 3 pixels wide?

Thanking you in advance. 

Best regards,

kaka21. 

[wiebe@CARYA]

Seems to me you're measuring in the margins. It would be fairly random what the result will be, purely based on how the exact pixels fall.

Any change you can get a higher resolution? That would be the best solutions.

I'd think a calliper\clamp tool would perform better, since the measurement would be based on a certain length, not just one line.

[crabman]

---

the gap is bigger in the image that has higher edge strength than the image that has a lower edge strength

---

I would not agree on this assumption. Are you definitively sure that the gaps are different? Because in my eyes this could be two pictures of the same gap where in one case the object is just slightly out of focus.

---

[kaka21]

Hello! 

Thanks for your response. I am considering to buy a high resolution optical system. For now, I have to work with the images I have. I have used the clamp tool, it seems to detect the contours as shown with the blue line, but is there any way to measure the dimensions of that contour? 

PFA the screenshots.

[kaka21]

Both the gaps are on the same image. It is the gap between the component and the deep drawing die during microforming process. For more details, please find the image in the attachments.