How to perform pixel-by-pixel matching to check whether a query image exist in the template database or not.

I am working on dorsal hand vein recognition system. I have already binarised and pre-processed the image followed by feature extraction (white pixels coordinates) of the thinned vein patterns as shown below in the figure (Image 1). These steps were repeated for 10 images and having their coordinates stored in .txt file.

http://imageshack.us/a/img19/2367/zsthin0073hv1.png

Now, let's say I have a query image (Image 2) as below where all the mentioned above steps have been applied and coordinates were retrieved.

http://imageshack.us/a/img543/9764/zsthin0059hv1.png

From here, how do I proceed with the Euclidean distance or pixel-by-pixel matching or any other technique to verify whether the query image exists in the database or not and also, generate a matching score?

Thank You.

Answer 1

[Anand]

Well, if you expect the images to be exactly the same, you can do this:

im1 = imread('http://imageshack.us/a/img19/2367/zsthin0073hv1.png');
im2 = imread('http://imageshack.us/a/img543/9764/zsthin0059hv1.png');
isequal(im1,im2)

If you don't think there's going to be am exact match, you could use the 2-D correlation coefficient using the corr2 function.

[Shirah]

Can you please provide me with a sample of code in which the corr2 function can be used for matching and generating a matching score. Thank You.

[Image Analyst]

Anand, do you really think this will work for the images they gave? Because I don't. A slight rotation will totally destroy any correlation. Since it's a skeleton, even using a jig to position the hand will not give subpixel accuracy and so the correlations will be almost zero.

Answer 2

[Image Analyst]

You might try to minimize the Hausdorf distance. See the bottom part of this page: http://cgm.cs.mcgill.ca/~godfried/teaching/cg-projects/98/normand/main.html or this: http://www.cs.cornell.edu/vision/hausdorff/hausmatch.html