im2uint16 makes images binary (to convert images from double to uint16)

I adjusted an image (uint16) using "imadjust" function.

The result image matrix is double, so I tried to convert it to uint16 using "im2uint16", but the result becomes binary (0 or 2^16-1).

How can I fix this?

My script is below, I am adjsuting a image stack.

% adjust images: stack0 (300x300x73 uint16)

range_in = [0 1]; %[0 1] for full range, [] for auto 1% saturation

range_out = []; %[] for "0-1" for imadjust output

stack_adjusted = nan(nx,ny,nfoi);

for f=foi

stack_adjusted(:,:,f) = imadjust(stack0(:,:,f),range_in,range_out);

end

%stack1=im2uint16((2^16-1)*stack_adjusted);

stack1=im2uint16(stack_adjusted);

if fInvertedImage

stack2 = imcomplement(stack1);

else

stack2 = stack1;

end

Answer 1

[DGM]

The image returned by imadjust() is of class 'double', but it's scaled as if it were still uint16. Im2uint16() will assume the image is normalized, given its class. When it denormalizes, all non-black content gets pushed out of the available range.

Normally, the output of imadjust() inherits its class (and scaling) from the input image. In this case, you've forced the output class to be 'double' in the preallocation using a NaN array. (NaN can't be represented in integer classes). You can use zeros() instead, and specify the class to match the expected output.

stack_adjusted = zeros(nx,ny,nfoi,class(stack0)); % preallocate
for f = 1:nfoi % i'm assuming you meant all frames?
    stack_adjusted(:,:,f) = imadjust(stack0(:,:,f),range_in,range_out);
end
% output doesn't need conversion anymore, since it's already uint16

[Myung Hyun Jo]

stack_adjusted = zeros(nx,ny,nfoi,'uint16');

zeros() also make it double, but I could specify the datatype and fixed the issue.

Thank you! DGM

[DGM]

If you use class(stack0) instead of explicitly specifying 'uint16', the preallocated array will be correct even when stack0 isn't uint16. I don't know if that will ever happen in your case.

[Myung Hyun Jo]

Got it. That sounds better. Thank you again!