AForge.NET

  :: AForge.NET Framework :: Articles :: Forums ::

Removing patterns by subtracting from "ground truth".

The forum is to discuss topics related to different areas of image processing and computer vision.

Removing patterns by subtracting from "ground truth".

Postby Avan Holzinger » Mon Oct 08, 2018 12:00 pm

Based on this thread I developed a new idea: as the vertical stripes look the same and are on the same position all over the screen, I copied the pattern and spread it all over the screen. My goal is now to compare all pixel of the current image of the video with pixels of the reference image and whenever the same pattern comes through, the background should be set to white.

Pixel by pixel comparison is bad, though, as it is about the pattern and not occasionally the same colors.

How do I do this in a performant way?
Avan Holzinger
 
Posts: 5
Joined: Fri Oct 05, 2018 10:22 am

Re: Removing patterns by subtracting from "ground truth".

Postby Avan Holzinger » Thu Oct 11, 2018 11:54 pm

HELLO? .... ELLO .... ELLO .... LLO

ANYONE HERE? ERE ERE ERE

Anyone hear me? me me me
Avan Holzinger
 
Posts: 5
Joined: Fri Oct 05, 2018 10:22 am

Re: Removing patterns by subtracting from "ground truth".

Postby andrew.kirillov » Fri Oct 12, 2018 7:22 am

What's up?

If you need pattern matching, then try searching for it, use Google. Pattern matching is not implemented in AForge.NET Framework, so can not recommend much.

If your pattern (lines, whatever) is always at the same location, then image subtraction could help or some other techniques.

When you say pixel by pixel is not fast, what do you mean? If you use something like GetPixel(), then forget about - it will never be fast. If you already use pointers and it is still not fast enough, then it looks like you are doing quite some heavy weight algorithm.
With best regards,
Andrew


Interested in supporting AForge.NET Framework?
User avatar
andrew.kirillov
Site Admin, AForge.NET Developer
 
Posts: 3327
Joined: Fri Jan 23, 2009 9:12 am
Location: UK




Return to Image Processing and Computer Vision