Filter out vertical pattern

[zou]

Please suggest the best way to remove the vertical pattern in the following image:

TIA

[Tanner!]

Can you give us some more information?

What is the ultimate goal?
What is this picture?
What data are you wanting?

If you can tell us more about your application we'll be better able to help you.

[zou]

> What is the ultimate goal?

> What is this picture?

> What data are you wanting?

Remove vertical pattern, so that when use IMAQ Find Edge, no false edge will be found.

This is a picture of a metal surface.

No vertical pattern should be found on this image.

[Tanner!]

Ok, why does your image have vertical lines? What camera are you using and what's causing that artifact?

Anything you do to filter out the vertical lines is going to affect your data. If you're looking to remove vertical noise from the image, taking the fft of the image and subtracting out the result could work. The lines don't look well spaced, and I think you'd be better off trying to acquire a better image rather than going down the rabbit hole of filtering techniques.

[zou]

It's no artifact.  The vertical patterns are real.  It's the scratch on the surface.  It's random spaced.

[BruceAmmons]

Perhaps a horizontal median filter which is wider than your scratches but not as wide as the features you are trying to detect.

There are so many scratches, I don't think much will be left after removing them.  If there was just a few bright scratches, it would be easier.

Bruce

[AnalogKid2DigitalMan]

Can you try more diffuse or cloudy day lighting methods?

-AK2DM

[zou]

Bruce,

Thanks for reply.

A median filter will remove the vertical pattern effectively.

However, it will also shift the real edge of object.

The edge is not visible in this shot, but will be in follow-up shots.

[zou]

---

@AnalogKid2DigitalMan wrote:

Can you try more diffuse or cloudy day lighting methods?

-AK2DM

---

Not sure how to do that.

Do you mean a Gaussian filter?

[AnalogKid2DigitalMan]

Not in software but the physical light source used to illuminate your specimen.

Lighting is often overlooked, it can make a great deal of difference in the signal to noise ratio in terms of highlighting features of interest or in minimizing features that are not of interest.

EDIT: Here is a good starting reference:

http://www.edmundoptics.com/technical-resources-center/illumination/choose-the-correct-illumination/

-AK2DM