Why the dimensions of my picture in pixels is reduced when I save my image?

Hi,

I filter an imageA with a function called bpass and then I display it with imagesc to be able to save it (as .tif). ImageA is 1038*1388 pixels, my filtered image is 1038*1388 double as well, but when I save it (right click on the image, save as...), the picture I then open is 560*420 pixels. Why...?

%Read the image
prepreim = imread('well8 aft.tif');
preim = rgb2gray(prepreim); % Convert uint8 color to uint8 gray scale.
im = double(preim); % Convert to double in the range 0-255.

%spatially filter the image
b = bpass(im,1,4);
figure
colormap gray
imagesc(b)

%save the filtered image

figure, colormap gray, imagesc(b); 
set(gca,'xtick',[],'ytick',[]); %remove axis
set(gca,'LooseInset',get(gca,'TightInset')); %remove white band

Not to save it manually, I tried to add the line

saveas(gcf,'8beffilter','tiffn');

But in this case I obtain a picture of dimension 875*656 pixels.

Thank you for your help.

Answer 1

[Image Analyst]

That probably saves the whole figure, not the image. Use imwrite() instead to save the image array only.

[Aude Rapet]

Thank you again Image Analyst, it works!!

[Aude Rapet]

The problem I have now is that my picture is not the same anymore when I call it with imshow. Before I used to do :

%Read the image
prepreim = imread('well8 aft.tif');
preim = rgb2gray(prepreim); % Convert uint8 color to uint8 gray scale.
im = double(preim); % Convert to double in the range 0-255.
%spatially filter the image
b = bpass(im,1,4);
figure
colormap gray
imagesc(b)

and I had this from imagesc:

and now that I have saved it with imwrite, when I open it later with imshow, I have :

Do you have an idea of what is wrong? Thank you

[Image Analyst]

imshow() normally doesn't apply a colormap while imagesc does. however it looks like you overrode the default crazy colored colormap that imagesc applied and made a gray scale one instead. Normally though, colormap() is called AFTER you call imagesc(), not before like you did, so I'm surprised that worked. So the only thing I can think of is that when the image is displayed with imshow() it's using some different intensity scale than when you displayed the same image with imagesc(). Or else you're displaying totally different images, not "b" by both of them. You're not showing the code where you called imshow(), only the code where you displayed it with imagesc(). Is the badly-named "b" a floating point array or a uint8 array? Can you save it to a .mat file so I can try displaying it with both functions to compare? Also, show your code for how you displayed "b" with both functions. If it's the same "b" then there must be a way to make it look the same way no matter whether you used imshow() or imagesc() to display it.

[Aude Rapet]

Here is my initial picture (before applying filter) but I use it in .tif nan .jpg:

My initial image (before filter) is a 1038x1388x3 uint8, then become 1038x1388 uint8 with rgb2gray (second line), then 1038x1388 double with imdouble (third line). b resulting from function bpass is 1038x1388 double.

Here is my code with imagesc and imwrite to save it :

%Read the image
prepreim = imread('well8 aft.tif');
preim = rgb2gray(prepreim); % Convert uint8 color to uint8 gray scale.
im = double(preim); % Convert to double in the range 0-255.
figure
colormap gray
imagesc(im)
whos im

%spatially filter the image
b = bpass(im,1,4);
figure
colormap gray
imagesc(b)

%save the filtered image 

figure, colormap gray, imagesc(b); 
set(gca,'xtick',[],'ytick',[]);
set(gca,'LooseInset',get(gca,'TightInset'));
imwrite(b,'afterfilter.tif');

Here is my second code where I use imshow :

%%STEP 1 : Read image 
original = imread('afterfilter.tif');   %without, t=1 AFTER
figure; imshow(original);

And I send you attached b.mat

Thank you

[Image Analyst]

You'll need to also give us code for bpass().

[Aude Rapet]

Here is the link to download the code for bpass : http://site.physics.georgetown.edu/matlab/code.html

Answer 2

[John BG]

if the bpass function is removing samples to low band pass filter, then the resulting image is going to be smaller.

have you checked that images b and im are same size?

John BG

<mailto:[e-mail] [e-mail]>

[Walter Roberson]

"my filtered image is 1038*1388 double as well"

[Aude Rapet]

Yes I checked :

>> whos im

Name         Size                 Bytes  Class     Attributes

    im        1038x1388            11525952  double

>> whos b

 Name         Size                 Bytes  Class     Attributes

b         1038x1388            11525952  double