How does Normalized Cross Correlation work?

I'm trying to understand how NCC works. For example I have used normxcorr2 for template matching. I understand how the size of the correlation image is configured. But I'm trying to figure out how the function works. How does the actual matching occur? And what is the correlation image used for?

Thank you very much for any information.

I have included the figures I have been working with as well as the code. .

.

.

.

Here is the code I have used:

%% Normalized Cross Correlation

 load('DataSet04.mat');
        TargetImage= imread('DataSet04-04.bmp');
        figure, imshow(TargetImage), title('Target Image for Matching (size  )'); 

        sampleTemplate2= zeros(168, 168);

 % Y, X1:X2 Top
 sampleTemplate2(5, 5:163) = 255;

 % Y, X1:X2 Bottom 

 sampleTemplate2(163, 5:163) = 255;

 % Y1:Y2, X  Right
 sampleTemplate2(5:163, 163) = 255;

 % Y1:Y2 X Left 
 sampleTemplate2(5:163, 5) = 255;

 sampleTemplate2 = imfill(sampleTemplate2, 'holes');
 figure, imshow(sampleTemplate2), title('Template used for Matching');
 imwrite(sampleTemplate2, 'tmplt2.bmp', 'bmp');
 tmplt2= imread('tmplt2.bmp');
       %figure, surf(sampleT2), shading flat;
        [Rowtmplt, Coltmplt] = size(tmplt2);

        CorrelationImage= normxcorr2(tmplt2,TargetImage);
        figure, imshow(CorrelationImage), title('Normalized Cross Correlation Image');
        [RowCorr, ColCorr] = size(CorrelationImage);
        sprintf('Size= %d%d', RowCorr, ColCorr);
        [maxCorr, imax] = max(abs(CorrelationImage(:))); 
        [ypeak, xpeak] = ind2sub(size(CorrelationImage), imax(1)); 
        sprintf('Row = %d%d', ypeak, xpeak); 

        BestRow = ypeak - (Rowtmplt - 1);
        BestCol = xpeak - (Coltmplt - 1);
        sprintf(' Best Match %d%d' , BestCol, BestRow);
        figure, imshow(TargetImage), title('Template Matched'); 
        h= imrect(gca, [ BestCol BestRow (Coltmplt - 1) (Rowtmplt - 1)]);

Thank You!

 % Amanda Johnson 

 % [e-mail]