Splitting image into 4 quadrants, up, down, right and left.

Hey! I am trying to split an image into 4 quadrants based on the two diagonals through the middle of the image. And then swap the upper quadrant with the lower one. I am having difficulty setting my loop to get the quadrants. I would really appreciate some help!! Thank you so much

Answer 1

[Matt J]

 [m,n]=size(yourImage);
 [I,J]=ndgrid(1:m,1:n);
 Jf=n+1-I;

 Q1=(I<J & J<Jf);
 Q2=(I<J & J>Jf);
 Q3=(I>J & J>Jf);
 Q4=(I>J & J<Jf);

 newImage=yourImage;
 newImage(Q3)=yourImage(Q1);  %flip top and bottom quadrants (Q1 and Q3)
 newImage(Q1)=yourImage(Q3);

[John L]

Hey Matt! Thank you for your answer! I wrote that down in matlab and ran it, I got this error message back

Unable to perform assignment because the left and right sides have a different number of elements.

Error in trial (line 13) newImage(Q3)=newImage(Q1); %flip top and bottom quadrants (Q1 and Q3)

[John L]

I am still getting the same error

[John L]

This is weird, I updated the code and I am still getting the same error

[Matt J]

Here is my test of the code.

yourImage = imread('cameraman.tif');

 [m,n]=size(yourImage);
 [I,J]=ndgrid(1:m,1:n);
 Jf=n+1-I;
 Q1=(I<J & J<Jf);
 Q2=(I<J & J>Jf);
 Q3=(I>J & J>Jf);
 Q4=(I>J & J<Jf);
 newImage=yourImage;
 newImage(Q3)=yourImage(Q1);  %flip top and bottom quadrants (Q1 and Q3)
 newImage(Q1)=yourImage(Q3);

 imshow(newImage)

[John L]

When I used the same image, it ran perfectly. But when I use a .png file, it gives me that error. Does it have to do with the size of the image, the one am using is 750x750. Thank you so much for taking time on this!

[Matt J]

Yes, probably. You have to be sure that the quadrants are the same size, in order to swap them.

[John L]

Do you have any suggestion what to do in this case?

[Image Analyst]

Make sure you have a gray scale image, not a color one. And if the number of rows and columns are not the same, use imresize().

Why do you want to do this unusual thing anyway?

[John L]

So I am given this image and the question says that the upper quadrant needs to be switched with the lower quadrant. I will make sure it is in grayscale first and then flip the quadrants.

Answer 2

[Image Analyst]

Here's a start (using some of Matt's code - thanks) that works for gray scale or RGB images, and non-square images (sort of). See if you can fix it (the Q's) so that the center of the diagonals is at the center of the final image for non-square images..

% originalImage = imread('cameraman.tif');
originalImage = imread('peppers.png');

[rows, ncolumns, numberOfColorChannels] = size(originalImage);
[Y,X]=ndgrid(1:rows,1:ncolumns);
Jf=ncolumns+1-Y;
Q1=(Y<X & X<Jf);
Q2=(Y<X & X>Jf);
Q3=(Y>X & X>Jf);
Q4=(Y>X & X<Jf);
flippedImage = flipud(originalImage);
% Mask out Q1 | Q3
% Mask the image using bsxfun() function to multiply the mask by each channel individually.
mask = Q1 | Q3;
maskedImage1 = bsxfun(@times, flippedImage, cast(mask, 'like', originalImage));
mask = Q2 | Q4;
maskedImage2 = bsxfun(@times, originalImage, cast(mask, 'like', originalImage));
finalImage = maskedImage1 + maskedImage2;
imshow(finalImage);