How to find a specific row and column RGB matrix

Hello. I want to select a RGB matrix from one image (For example as rows-200 to 220 and coloumn 230to246 matrix) and want to find pixel location of same matrix in another image. For that I wrote below program.Here I1 is first image,I2 is second image and Irefmat is specific matrix from image I1 which I want to find in I2.

Irefmat=I1(100:120,100:120);

red=I2(:,:,1);

green=I2(:,:,2);

blue=I2(:,:,3);

mask=red == Irefmat

green == Irefmat

blue == Irefmat ;

[rows,columns]=find(mask);

subplot(2,2,1);

imshow(mask);

But error is coming. Looking forward for some help!!!

[DGM]

You're trying to do direct comparison between arrays of different size. That doesn't work.

mask=red == Irefmat

What you're looking for is basic pattern matching. Testing for equality is going to fail unless we're dealing with an exact integer copy of the pattern sample.

A naive solution would be a sliding window difference filter. The result from such a filter would be zero (or approximately zero) only when a pattern match occurs. Such a filter would not be able to detect transformed (scaled, rotated) instances of the pattern.

I'm pretty sure Matlab has more robust tools for doing this (I'm not familiar, if there are). Consider these answers.

https://www.mathworks.com/matlabcentral/answers/24467-identifying-objects-in-a-picture-containing-several-objects

This is from the CVT docs:

https://www.mathworks.com/help/vision/ug/pattern-matching.html?searchHighlight=pattern%20matching&s_tid=srchtitle

[Constantino Carlos Reyes-Aldasoro]

IT would be easier to help if you illustrate your problem with actual images.

[Arpit Bhatt]

i have attached two images.I want to take one pixel matrix from image1 and want to find same matrix in image2 so that I can find how much spatial deviation has happened.Any other methods are also welcomed!!!

Answer 1

[Image Analyst]

I do exactly that in my demo of normxcorr2(). I extract out a rectangular ROI (white onion) from an image (that contains the onion) and then find it in the image. I use the same image, but you could just as well find it in a different image.

[Arpit Bhatt]

Thank you for your reply. Your code workes fine. I am having problem in furthur extension. Actually I am using manual selection of matrix and finding the same matrix in second image. The goal is to find displacement of matrix in terms of rows & columns. The problem is after having xpeak and ypeak, if i compare the results than it gives wrong value. I am sharing two images in which second image is delibrately shifted one row and one column. So the values which the corelation output gives should be nearbz, which is not. Any idea regarding process is welcomed.

smallSubImage = imcrop(rgbImage, [m, n, sampleWidth, sampleHeight]);

% Get the dimensions of the image. numberOfColorBands should be = 3.

[rows, columns, numberOfColorBands] = size(smallSubImage);

subplot(2, 2, 2);

imshow(smallSubImage, []);

set(gcf,'MenuBar','none')

axis on;

caption = sprintf('Sample to Search For, %d rows by %d columns.', rows, columns);

title(caption, 'FontSize', fontSize);

channelToCorrelate = 1; % Use the red channel.

tic

correlationOutput = normxcorr2(smallSubImage(:,:,1), rgbImage(:,:, channelToCorrelate));

toc

subplot(2, 2, 3);

imshow(correlationOutput, []);

axis on;

% Get the dimensions of the image. numberOfColorBands should be = 1.

[rows, columns, numberOfColorBands] = size(correlationOutput);

caption = sprintf('Normalized Cross Correlation Output, %d rows by %d columns.', rows, columns);

title(caption, 'FontSize', fontSize);

% Find out where the normalized cross correlation image is brightest.

[maxCorrValue, maxIndex] = max(abs(correlationOutput(:)));

[yPeak, xPeak] = ind2sub(size(correlationOutput),maxIndex(1))

% Because cross correlation increases the size of the image,

% we need to shift back to find out where it would be in the original image.

corr_offset = [(xPeak-size(smallSubImage,2)/2) (yPeak-size(smallSubImage,1)/2)]

% Plot it over the original image.

subplot(2, 2, 4); % Re-display image in lower right.

imshow(rgbImage2);

axis on; % Show tick marks giving pixels

hold on; % Don't allow rectangle to blow away image.

% Calculate the rectangle for the template box. Rect = [xLeft, yTop, widthInColumns, heightInRows]

boxRect = [corr_offset(1) corr_offset(2) sampleWidth, sampleHeight]

% Plot the box over the image.

rectangle('position', boxRect, 'edgecolor', 'g', 'linewidth',2);

% Give a caption above the image.

title('Sample area Found in Second Image', 'FontSize', fontSize);

% Display the original color image.

% Plot it over the original image.

subplot(2, 2, 1); % Re-display image in lower right.

imshow(rgbImage);

axis on; % Show tick marks giving pixels

hold on; % Don't allow rectangle to blow away image.

% Calculate the rectangle for the template box. Rect = [xLeft, yTop, widthInColumns, heightInRows]

boxRect = [corr_offset(1) corr_offset(2) sampleWidth, sampleHeight]

% Plot the box over the image.

rectangle('position', boxRect, 'edgecolor', 'g', 'linewidth',2);

% Give a caption above the image.

title('Sample area in First Image', 'FontSize', fontSize);