Computer Science Theory: Image Similarity

Question

So, I'm trying to compare different images and wondered if someone could point me in the right direction with some basic metrics that I can take for a group of images.

Assuming I have two images: A and B, I pretty much want as much data as possible about each, so I can later compare them programmatically. Things like “overall color,” “overall shape,” etc., would be great.

If you can help me find specific properties and algorithms to calculate them, it will be great!

Thank!

EDIT: The ultimate goal here is to be able to tell the computer how “similar" images are too. If two images are the same, but one has a blurred face; they should be recorded as fairly similar. If the two images are completely different from each other, the computer should be able to say.

Answer 1

What you are talking about is much more general and non-specific.

Image information is formalized as Entropy .

What you seem to be looking for is basically extracting the functions and then comparing these functions. There are many possibilities that can be learned, but many of them may not be relevant depending on differences in images.

, . , 100 , .

, , , .

, .

()

, , - , . , , .

, , , , , , , , :

1) ( ): . * height - . .

2) . ( FTT , OpenCV), 2D. , , , , . . , .

Answer 2

SURF -

( Mathematica), (, ):

Answer 3

. ...

Answer 4

OpenCV. ( ) .