[Vision] Color Pattern Matching and Pattern Matching

[lord_nox]

Hello again guys, this question is more theoretical than practical: I want to know what's the difference between "Color Pattern Matching" and "Pattern Matching", I've tried to find something online, but I can't see any difference between the two options of vision recognition.

I'd appreciate a lot if someone could send me a link or an explanation of this.

Thanks in advance!

[wiebe@CARYA]

https://zone.ni.com/reference/en-XX/help/370281AC-01/nivisionconcepts/color_pattern_matching/

First section (When to use) begins with grayscale pattern matching.

http://zone.ni.com/reference/en-XX/help/372916P-01/nivisionconcepts/pattern_matching_techniques/

The obvious difference is one works with color images, the other does not...

[Bob_Schor]

It basically comes down to a Math problem.  Suppose I have a GrayScale Image, and am examining a Pixel and its neighbors, and want a measure of how near they are in value.  The simple way to do this is to subtract the two GrayScale values, which typically are 8-bit unsigned integers, and to ask how close the (signed) difference is to 0.

Now do this with a Color Image.  The Pixel values are now points in a 3-dimensional "Color Space", and can be expressed in terms of RGB 8-bit values or Hue-Intensity-Saturation 8-bit values, and the concept of "distance" (or "how similar are these two pixels") is much more complicated and, potentially, subjective.

For this reason, many Machine Vision routines (where "Color" is not the salient feature, such as "Isolate the Red Marbles") start with "Convert the image to GrayScale".

Bob Schor