Ask Your Question

Texture Matching using LBP/H? C++

asked 2015-10-03 11:34:52 -0500

Leo_ gravatar image

updated 2015-10-06 08:46:57 -0500

I have a question about LBPs, but firstly let me explain what I am trying to do.

I am trying to match coins using LBPs. I have an input coin and I want to compare it with other coins and see if the coin is of the same denomination. Attached below are 3 images:

  • R1 - the input image - the one to be compared to
  • R1T - needs to be compared to R1
  • R2T - needs to be compared to R1

I have found a very good source here: Local Binary Patterns

All my code needs to simply do is say that R1T is a match and R2T is not a match.

How can I implement something like this?


Images: Thumbnails

edit retag flag offensive close merge delete


I think going with a texture matching technique would be an overkill. I think that it would be better if you were trying first to extract the arithmetic information from the images by using an OCR approach and then determine a match by comparing this information for each sample.

theodore gravatar imagetheodore ( 2015-10-03 11:56:54 -0500 )edit

@theodore there are some coins without text or numbers. So I need to do texture matching.

Leo_ gravatar imageLeo_ ( 2015-10-03 12:02:52 -0500 )edit

@Leo_ ok I see. You did not mention that on your initial message so I thought that all your images would contain numeric information.

theodore gravatar imagetheodore ( 2015-10-03 12:09:43 -0500 )edit

1 answer

Sort by ยป oldest newest most voted

answered 2015-10-05 07:51:03 -0500

ThomCastel gravatar image

If your images are all like the ones you showed us, meaning no scale or illumination change and nice background, then you could either use:

  1. Template matching using matchTemplate and select the sample that gives the best result.
  2. Binarize, get blobs and sort the coins based on shape or area.

That's what I would try first, but maybe you have some other constraints?

edit flag offensive delete link more

Question Tools

1 follower


Asked: 2015-10-03 11:34:52 -0500

Seen: 1,139 times

Last updated: Oct 03 '15