How to differentiate between two attached images with 25 sand particles in each image. I have successfully calculated circularity of particles in images. Average value is similar for both samples. Please guide to how differentiate between two sands?

% The sand1 and sand2 images used are attached. In order to remove the noise in images median filter will variable rectangle window is used.
% for sand1 bw = medfilt2(bw, [30 30]); 
% for sand2 bw = medfilt2(bw, [12 12]);
% rectangle threshold is fixed when the noise just disappears.
% An important to note is that with change in this window the circularity changes. How to deal with this ???????



clear all;
close all;
clc;

RGB = imread('sand1.jpg');
I  = rgb2gray(RGB);
bw = imbinarize(I);
bw = bwareaopen(bw, 35);
se = strel('disk', 2);
bw = imclose(bw, se);
bw = imfill(bw, 'holes');
bw = medfilt2(bw, [12 12]);
[B, L] = bwboundaries(bw, 'noholes');
numberOfBoundaries = size(B, 1);
imshow(bw);
hold on;
for k = 1 : numberOfBoundaries
                 thisBoundary = B{k};    
                 plot(thisBoundary(:,2), thisBoundary(:,1), 'R', 'LineWidth', 2);
                 hold on;
end
hold off;

stats = regionprops(L, 'Area');

for k = 1:length(B)
                  thisBoundary = B{k};
                  % computing perimeter
                  delta_sq = diff(thisBoundary).^2;    
                  perimeter = sum(sqrt(sum(delta_sq,2)));
                  area = stats(k).Area;
                  % computing circularity
                  circularity(k) = 4*pi*area/perimeter^2;
end 

[KALYAN ACHARJYA]

Attched images?

Please be clear, what you have, what result you are expecting?

[Sumit Chaudhary]

Yes, I have attached the images.

I have two different types of sand particles, which needs to be differentiated?

I have taken 25 particles out of the sand and taken the images. Now, the obtained average circularity values for 25 particles for two different sand is similair i.e 0.79 and 0.78. The calculated circularity values vary with noise filtering window of median filter. Is this affecting my true circularity? So, I was wondering what could i do more to differentiate between the sands ?

[KALYAN ACHARJYA]

Please correct me, If I misundestood the question.

You want to check wheather the both image are same or not (simmilarity)?

[Sumit Chaudhary]

This two are sands from two different places having different physical and chemical property. Therefore, we are looking for parameters through image processing by which we can differentiate between these sands based on their shapes. In the above code, I have calculated circularity (shape parameters that gives how close are a shape is to a cirlce value and the value of circularity is 1 for perfect circle) but the average values is similair for these to sand.

Are we on the same page brother ?