Pattern matching for binary images

[JulianC]

Hi, 

I'dl like to know if pattern matching works in the same way between a grayscale (U8) image and a binary image (after applying extract single color plane, IVA gray filters Nth order, grayscale threshold and IVA binary inverse filters).

I'm trying to have reasonable results because I'm getting a score of 1000 for searching the templates in the same image from which they were created, which is expected, but when I take a picture with my GigE cam and apply all those filters with the templates previously created I get scores around 700.

Julian

[Guillermo.O]

Hello JulianC,

Answering your question, the grayscale pattern matching works a little different with a binary image, this is because the grayscale method uses information based only of textures and shadows in the object or pattern you are searching for, and with binary pictures you don't have this information, so you can only detect edges of your pattern.
About the test you did with the pattern matching, it is normal that with the original picture where you get the pattern you will get a 1000 score (Max. possible), and when you are searching for the pattern in other picture or directly with a camera, it is normal that you found patterns with lower score, the reason of this results may be the changes with your lighting setup, or also some noise in your camera. Still I can tell you 700 is a good score for pattern matching, you can change the minimum score for the detection of patterns.

Hope this information will be useful for you.

Greetings

Guillermo O.

National Instruments