Assign values from an matrix to another by using row and column indexes -- array exceeds maximum array size preference

Hi

So I have an image (345*345 matrix), and I did this

[row, column] = find(image == 1);

i.e., to get the row and column indexes for all pixel values that is 1.

This is a binary image, so a lot if pixles have value 1. So the 'row' and 'column' each returns a 57033*1 size big array.

Now I want to assign a different values from another matrix to those pixel values that originally were 1. And they should be 1-1 matching from the two images using those indexes, something like below:

image(row,column) = image_new(row, column);

But then matlab returned an error:

Requested 57033x57033 (24.2GB) array exceeds maximum array size preference (16.0GB). This might cause MATLAB to become unresponsive.

Just wondering is there still a way to do what I propose to do. i.e., assign the values that were 1 from a new matrix in the same pixel positions? (you can think that the new matrix was like a foreground of the image to fill into the original binary image.)

Thanks in advance!

Answer 1

[Image Analyst]

You can just use the logical image. Don't use find or rows or columns at all. Simply do

mask = image1 == 1; % Logical image.
image1(mask) = image_new(mask)

[Dyuman Joshi]

Logical values only take 1 byte per element, whereas double values take 8 byte per element. So the total memory costs will be cut by 7/8th

Not only that, logical indexing is much more faster than find().