Cross Correlation Map for stack of images

Hello guys,

I am wondering if someone here could help me with correlation map.

Basically I have some stacks of images and I would like to cross correlate them using the Pearson

Correlation coefficient formula below.

Ai and Bi are the intensity values for a particular pixel in the pattern, Abar and Bbar are the

mean values of the intensities and sigmaA and sigmaB are the standard deviations for the two CXD

patterns, A and B. The sum is evaluated over the 3D stack of images.

Anything would be nice. I have ran out of clues to implement this properly.

Basically in my script I have the following, sending to my pearson correlation function the stack of pixels:

for i=1:size(crop_cell,2)%Running over the cell stacks
    for j=1:size(crop_cell,2)%Running over the cell stacks
        for k=1:size(crop_cell{1},1)%Running over the rows
            for L=1:size(crop_cell{1},2)%Running over the columns
                corr_matrix{i,j}(k,L) = pearson_corr_scores_pix(crop_cell{i}(k,L,:),crop_cell{j}(k,L,:));
            end
        end
    end
end

And the correlation function:

function [p] = pearson_corr_scores_pix(a,b)
mean_a = mean(a);
mean_b = mean(b);
fac_a = a - mean_a;
fac_b = b - mean_b;
std_a = std2(a);
std_b = std2(b);
score_a = fac_a/std_a;
score_b = fac_b/std_b;
p = (1/(length(a)-1))*sum(score_a.*score_b);
end

Attached is my result(first) and a reference result (second) I would like to reach.

Mine is somehow not taking into account the correlation with all the stack of pixels

in the image? I don't know.

Thanks

Answer 1

[Subhadeep Koley]

Find the attached Pearson Correlation Coefficient function myPCC.m implemented in terms of covariance according to the documentation.

Run the following script to get the PCC Map.

close all;clc; 
% read all images from your folder and making them of same dimension 
myFolder = 'your_path'; 
if ~isfolder(myFolder) 
  errorMessage = sprintf('Error: The following folder does not exist:\n%s', myFolder); 
  uiwait(warndlg(errorMessage)); 
  return; 
end 
filePattern = fullfile(myFolder, '*.tiff');% change the extension according to your file extension 
tiffFiles = dir(filePattern); 
for k = 1:length(tiffFiles) 
  baseFileName = tiffFiles(k).name; 
  fullFileName = fullfile(myFolder, baseFileName); 
  fprintf(1, 'Now reading %s\n', fullFileName); 
  temp = double(imread(fullFileName)); 
  [r,c,d]=size(temp); 
  if d==3 
     temp=rgb2gray(temp); 
  end 
  if r~=256 && c~=256 
     temp=imresize(temp,[256,256],'bicubic'); 
  end 
  imageArray{k} =temp; 
end 
% calculate PCC for every image combinations 
for i=1:size(imageArray,2) 
    for j=1:size(imageArray,2) 
        corr_cell(i,j)=myPCC(imageArray{i},imageArray{j}); 
    end 
end 
% plotting the Pearson Cross Correlation Coefficient Matrix 
imagesc(rot90(flipud(corr_cell))); 
colormap(parula); 
set(gca,'YDir','normal'); 
set(gca, 'XTick', 1:2:length(tiffFiles)); 
set(gca, 'YTick', 1:2:length(tiffFiles)); 
title('Pearson Cross Correlation Coefficient Matrix'); 
colorbar; axis square;