null or NaN entries in image arrays?

Is it possible to set some elements in animage array to null or in some way that they are ignored?

I have an automatically detected circular ROI outside of which is to be ignored (those values will skew the results) and need to step through the array using a for loop. Obviously a for loop can only step in rectangles so is there any way I can set an element in that image array to null or NaN that is acceptable?

Regards

Tim

[Manikanta Aditya]

Yes, you can set elements in an image array to NaN to effectively ignore them in your calculations.

1. Create a mask for the circular ROI.
2. Apply the mask to the image array to set the values outside the ROI to NaN.

[Tim]

Hi Manikanta, thanks for the reply. I tried manually setting an element in the resultant array to NaN but it set it to 0 which isn't the same thing. Maybe I should try programetically?

[Manikanta Aditya]

Yes, you can programmatically set elements in an image array to NaN to ignore them in your calculations.

[DGM]

What is the actual applied operation? Does it actually require a loop, or can it be done with basic logical indexing?

Answer 1

[Epsilon]

Hi Tim,

It is possible to set the elements to NAN in order to be ignored. Attaching an example:

% Sample image array
img = rand(100);

% Define the circular ROI
[rows, cols] = size(img);
center = [50, 50];
radius = 20;

% Create a mask for the circular ROI
[x, y] = meshgrid(1:cols, 1:rows);
mask = sqrt((x - center(1)).^2 + (y - center(2)).^2) <= radius;

% Set elements outside the circle to NaN
imgModified = img;
imgModified(~mask) = NaN;

% Display original and modified images
figure;
subplot(2, 1, 1); imagesc(img); title('Original'); colorbar;
subplot(2, 1, 2); imagesc(imgModified); title('Modified'); colorbar;

Hope it helps.

[Tim]

That worked well. Thanks Epsilon

[Walter Roberson]

This code relies upon img being type double or single . This code would not work if img is uint8() for example. You really should use im2double() like I showed in my Answer.

Answer 2

[Walter Roberson]

The trick is that if your image array is not datatype single() or double(), then there is no way to set locations to be ignored as part of the same array

You can, however,

Imd = im2double(YourImageArray);
Imd(AppropriateLocations) = nan;

This will not in itself cause locations to be skipped. For example if you have

for ROW = 1 : size(YourImageArray,1)
    for COL = 1 : size(YourImageArray,2)
        do something with Imd(ROW,COL)
    end
end

then this code in itself will not skip the locations that are NaN. You would need

for ROW = 1 : size(YourImageArray,1)
    for COL = 1 : size(YourImageArray,2)
        if ~isnan(Imd(ROW,COL))
            do something with Imd(ROW,COL)
        end
    end
end

but if you are going to do that then you might as well use

for ROW = 1 : size(YourImageArray,1)
    for COL = 1 : size(YourImageArray,2)
        if YourImageMask(ROW,COL)
            do something with YourImageArray(ROW,COL)
        end
    end
end

Answer 3

[Steven Lord]

If your images are stored in an integer data type, assigning a NaN value into that array will store a 0 in that location. There is no equivalent of a NaN value in the integer types; all the possible bit patterns that can be stored in that type represent finite values.

A = int16(magic(3))

A = 3×3

8 1 6 3 5 7 4 9 2

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

A(3, 3) = NaN % 2 replaced by 0 not NaN

A = 3×3

8 1 6 3 5 7 4 9 0

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

B = uint32(magic(4))

B = 4×4

16 2 3 13 5 11 10 8 9 7 6 12 4 14 15 1

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

B(2, 4) = NaN % 8 replaced by 0 not NaN

B = 4×4

16 2 3 13 5 11 10 0 9 7 6 12 4 14 15 1

<mw-icon class=""></mw-icon>

<mw-icon class=""></mw-icon>

Answer 4

[Tim]

Thanks for your help guys.