only masking part of image help

Is there a way to apply an image mask on only part of the image?

When I use data tips, I was able to find the dimensions of where exactly I want my mask to appy (x and y values) and I tried to do something like this

frame(frame(755:933,206:388)<threshold)=0

but I recieved an error saying that the numbers were out of bound.

Answer 1

[DGM]

There's at least one thing going on here. This operation

frame(755:933,206:388)<threshold

will return a 179x183 logical image, which is smaller than the source image. This is going to cause an indexing problem -- not necessarily an error, but it won't be addressing the same part of the image. You're doing two indexing operations here. One based on the ROI location, and one based on the logical comparison. You need to keep track of both. This is a simple way:

% a simple image
A = imread('cameraman.tif');

% coordinates of ROI
x = [50 206];
y = [50 100];

threshold = 180;

Aroi = A(y(1):y(2),x(1):x(2)); % extract ROI
Aroi(Aroi<threshold) = 0; % operate on it
A(y(1):y(2),x(1):x(2)) = Aroi; % insert it back into the image

imshow(A)

That said, the above issue shouldn't be causing an out-of-bounds error since the ROI is smaller than the parent image. If you're getting an error like that, you might want to make sure that your subscripts are in the right order (y,x instead of x,y).

[Nara Sixty Five]

Thanks so much for the help. I was using the order x,y rather than y,x like you mentioned!

Answer 2

[Image Analyst]

An alternative way is:

frame = imread('concordorthophoto.png');% Read in sample image.
threshold = 110; % Whatever you want.
mask = frame < threshold; % Get binary image.
% Assign binary image in the region to the original image in the same region.
frame(755:933, 206:388) = 255 * mask(755:933, 206:388);
imshow(frame); % Display image with binary portion in left side.
axis('on', 'image'); % Show tick labels for rows and columns.

Note the binarized part on the left side of this image between rows 755 and 933, and columns 206 to 388.

[DGM]

Instead of binarizing the ROI

frame(755:933, 206:388) = 255 * mask(755:933, 206:388);

just need to set the dark fraction of the region to black

frame(755:933, 206:388) = frame(755:933, 206:388) .* uint8(~mask(755:933, 206:388));

[Image Analyst]

Ah, you're right @DGM (I didn't read closely enough). Thanks for the correction.