Name for a test to find the best match for a given image in a library of images (OpenCV?)

Question

I have looked through all the OpenCV documentation that I could find and googled for what I'm looking for, but I think I must be using the wrong terminology.

What I want to do is something similar to template matching, but I do not need to search for instances of my image in another image. I need to compare my image to a library of other images and determine which is the best match. Is there a name for this kind of matching?

If there is no existing library for it, I am planning just to use a correlation function for each image in the library and take the best match.

Answer 1

It sounds like image matching or image comparison (another keyword you might search for is image registration but that is a bit different).

Answer 2

Try looking for SIFT/SURF.

Answer 3

One simple technique could be to compare the histograms of the two images. Check out this link for more information

http://opencv.willowgarage.com/documentation/cpp/histograms.html?highlight=histograms#cv-comparehist

You should also look at the sample programs that come with openCV called find_obj and find_obj_ferns

Answer 4

Well you want to match a template in a lot of images.

You can create a loop for matching teplates to images. And find minMaxLoc and values.

So you can see which image how match to template.