Image Processing Toolbox: Subtracting 2 images

Hi everyone! I am working with an image of a cone. Attached:

As you can see, there is a lot of imperfections and I will like to remove them and leave only the black outline of the cone. My first thought was to take the same picture, but without a cone and the substract them. I attach the second image without the cone.

Nevertheless, the operation isnt working as I will expect. The results of the operation are the following, each accounting for different substractions.

In this image imperfections are still present.

And in this image I tried increasing the brightness but didnt work.

Am I thinking the method wrong? Or is there any other better method to process this images?

Thank you in advance!

Answer 1

[Image Analyst]

Try this. I trust you can find the angle yourself now that I've found the equation of the lines:

clc;    % Clear the command window.
close all;  % Close all figures (except those of imtool.)
clear;  % Erase all existing variables. Or clearvars if you want.
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 20;

%--------------------------------------------------------------------------------------------------------
%    READ IN IMAGES
folder = pwd;
baseFileName = 'ConeReference.png';
image1 = imread(baseFileName);
baseFileName = 'ConeTest.png';
image2 = imread(baseFileName);
% Get the dimensions of the image.
% numberOfColorChannels should be = 1 for a gray scale image, and 3 for an RGB color image.
[rows, columns, numberOfColorChannels] = size(image1)
if numberOfColorChannels > 1
	% It's not really gray scale like we expected - it's color.
	% Use weighted sum of ALL channels to create a gray scale image.
	image1 = min(image1, [], 3);
end

%--------------------------------------------------------------------------------------------------------
% Display the images.
nexttile
imshow(image1, []);
axis('on', 'image');
title('Reference Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;
hFig = gcf;
hFig.WindowState = 'maximized'; % May not work in earlier versions of MATLAB.
% Now show image 2
nexttile
imshow(image2, []);
axis('on', 'image');
title('Test Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;
drawnow;

% Get a mask of the bright parts in image 2.
% imhist(image1);
% grid on;
mask = imbinarize(image2);
nexttile
imshow(mask, []);
axis('on', 'image');
title('Mask Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;
drawnow;

% Get the means of each image within the mask region.
meanRef = mean(image1(mask))
meanTest = mean(image2(mask))
% Scale image 2 so that the mean in the bright region will be the same as for image1.
% This will enable better, more complete subtraction.
image2 = uint8(double(image2) * meanRef / meanTest);
% Double check the mean - should be really close as image 1 now.
meanTest = mean(image2(mask))

% Compute the difference image.
diffImage = imabsdiff(image1, image2);
% Display the image.
nexttile
imshow(diffImage, []);
axis('on', 'image');
title('Difference Image', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;

% Check the mean of the subtraction image in the mask region - should be close to 0.
meanSubtraction = mean(diffImage(mask))

% Threshold to find the cone.
coneMask = imbinarize(diffImage);
nexttile
imshow(coneMask, []);
axis('on', 'image');
title('Cone Alone', 'FontSize', fontSize, 'Interpreter', 'None');
impixelinfo;
drawnow;

% Find the highest y value - the top of the cone
[r, c] = find(coneMask);
[minR, indexOfMinR] = min(r)
peakC = c(indexOfMinR);
hold on;
xline(peakC, 'Color', 'r', 'LineWidth', 2); % Put a line there.
% Get left half coordinates
leftx = c(c < peakC);
lefty = r(c < peakC);
% Fit to a line
coefficientsLeft = polyfit(leftx, lefty, 1)
yFittedLeft = polyval(coefficientsLeft, leftx);
plot(leftx, yFittedLeft, 'r-', 'LineWidth', 2); % Overlay fitted line over image.

% Get right half coordinates
rightx = c(c >= peakC);
righty = r(c >= peakC);
% Fit to a line
coefficientsRight = polyfit(rightx, righty, 1)
yFittedRight = polyval(coefficientsRight, rightx);
plot(rightx, yFittedRight, 'r-', 'LineWidth', 2); % Overlay fitted line over image.

msgbox('Done');

[Urbano Alfonso Medina Martinez]

Worked perfectly!

Can´t thank you enough.

Answer 2

[Image Analyst]

What I'd do is to make a template of the bright spot without the cone. Then determine the means and scale them, then subtract.

% Get the mean of each image.
mean1 = mean(refimage(mask))
mean2 = mean(testImage(mask));
% Scale test image's mean to be the same as the reference image's mean.
testImage = double(testImage) * mean1 / mean2;
% Cast it back to the same image type as refImage.
testImage = cast(testImage, 'like', refimage);
% Now subtract.
subtractionImage = imabsdiff(refImage, testImage);

Now they should subtract much closer to zero than with what you did.

[Urbano Alfonso Medina Martinez]

Im really sorry but I wasn´t able to run the first line of code. I apologize, I am new to image processing.

The problem I am having is how to define the mask. So, applying a mask to a matrix (which corresponds in this case to the image) is extracting all those values which obeys the mask condition, i.e., numbers greater or smaller than the mask. Then I will simply find the mean of this values by using the mean function.

But how do I define an appropriate mask?

That is my question. Thanks!