My data type changes to uint8 after if condition I want to keep it int16

I'm running a for loop with an if condition. In this loop I apply conditions to an int16 image after reading it. However, for some reason the data type changes to uint8 which I don't want. I want to keep the data type the same (int16). Help!!!!

Here is the code I'm using:

%I want to keep data type the same!! 
for TrialFrame=1:num_stackimages;            
       image=imread(fullFileName,TrialFrame); %Here the data type is int16

        if Trials==1
        SubVideoGREEN(:,:,TrialFrame)=image; %Here data type changes to uint8
        else 
        TrialFrame2=TrialFrame+num_stackimages.*(Trials-1);        
        SubVideoGREEN(:,:,TrialFrame2)=image;  %Here data type changes to uint8      
end
   end

[Bjorn Gustavsson]

First a minor suggestion. Avoid (at all costs) shadowing built-in functions by pre-using their name for variables. My suggestion is that you should rename your variable image to curr_img to avoid shadowing the built-in function image.

[Stephen23]

The class of (very badly-named) image is irrelevant, all that matters is the class of the LHS array.

Did you preallocate SubVideoGREEN as uint16 ? Please show us the code where it is defined.

[Joseph Cheng]

if the LHS (left hand side) isn't previously defined check to see the image isn't actually uint8 image. This is because i'm seeing that you're expecting the imread of the image to be in int16(?). I don't recall a typical format image file that imread() can read in as int16. the documentation shows uint16 for the different formats that you can specify the frame to be read. for a sanity check you should look at the imfinfo() of the image file to actually verify that you're looking at the bit depth you're expecting and not the uint8 that i'm expecting your actual file to be.

[dpb]

And, what difference does it make for internal storage, anyways? Use what MATLAB defaults to; if at some point there is a need to cast to another class, then do it then.

But, you've given no indication of why it matters at all here...

[Bjorn Gustavsson]

@dpb, well the line:

TrialFrame2 = TrialFrame + num_stackimages .* ( Trials - 1 );

Might be tricky in uint8...

[dpb]

But that's not the variable complained over...it'll be a double unless it's been cast somewhere earlier...which, of course, without rest of code and without the data files it's pretty-much a guess as to why/what is happening...

[Lina Koronfel]

@Bjorn Gustavsson,@Stephen Cobeldick,@Joseph Cheng,@dpbThank you so much for the replies. I did preallocate SubVideoGREEN as uint8 which was way earlier in the main code so I missed it (my bad!!). However, as you recommedned, if the type of variable is not identified, it will result in a double SubVideoGREEN matrix . Thus it is important to identify it as 'int16' when preallocating. I also changed 'image' to something else. Now it is working fine.

@dpb The bit size matters because what this code does is concatenating a stack of images into a bigger stack, which I convert back into images (frames), and I don't want to lose the original information, which is recorded as int16.

Here is the code as it works now. Thank you!!

SubVideoGREEN=zeros(H,W,F,'int16');

   for TrialFrame=1:num_stackimages;            
       FrameinStack=imread(fullFileName,TrialFrame); %reads int16

        if Trials==1
        SubVideoGREEN(:,:,TrialFrame)=FrameinStack; %now allocated as int16
        else 
        TrialFrame2=TrialFrame+num_stackimages.*(Trials-1);        
        SubVideoGREEN(:,:,TrialFrame2)=FrameinStack; %now allocated as int16       
end
   end

Answer 1

[Jan]

A simplification of your code:

SubVideoGREEN = zeros(H, W, F, 'int16');
for TrialFrame=1:num_stackimages;            
    TrialFrame = TrialFrame + num_stackimages * (Trials - 1);        
    SubVideoGREEN(:, :, TrialFrame) = ...
        imread(fullFileName, TrialFrame);
end

There is no need to distinguish the case Trial==1 from the other frames, because the formula to get the TrialFrame replies the same value for both cases.