Why does MATLAB play with the contrast of my image after saving it?

I have a very weird problem that I hope someone can help me with. So Attached are two pictures of almost the exact same location in a map of a desert, one named "true" and one named "weird". The "true" image is the one I have and downloaded directly from the source in a tiff format. And the "weird" image is the true image taken into MATLAB with the following piece of code:

pictureData = imread("true.tiff");
h = imshow(pictureData);
imwrite(h.CData,"weird.tiff")

Now I don't understand why my map's colors are very slightly adjusted after saving the new image where the new image has lighter colors (the colors are almost the same I know...but still there is a noticeable difference that will skew my results)

I would like to do some analysis on my images but I am afraid that MATLAB will ruin my pictures and skew my results with this arbitrary change in colors. Is this because my images are gigantic (>50MB)? Or is it because I am saving those images incorrectly? I don't really know whats going.

[Bjorn Gustavsson]

The images you've attached are not of the same size. For anyone to be able to help you figure out what's the cause, I suggest you attach images that are of the same size and region of the original images.

[Walter Roberson]

I would check whether the original image has a color map. I would also check if the original image is 16 bit.

[Ali Almakhmari]

The original image has values that range from 0 to 255 and of type uint8.

[Walter Roberson]

Your code shows .tiff but the posted images are .jpeg . .jpeg is lossy and should not be used for scientific purposes (at least not without the lossy compression being specifically taken into account.)

[Ali Almakhmari]

yeah that cause I took screen shots of the images. Do you want me to send a link of the full images?

[Walter Roberson]

Yes, actual images would be best for testing.

[Ali Almakhmari]

Sorry for the delay. I uploaded the full images here. You can see that their is a very slight change in the contrast between the "true" and "fake" one. Even try taking the "true" one to MATLAB and saving it the same way I did to see how it changes.

https://drive.google.com/drive/folders/1FOcLAqg2Rw1AWfzJ3bkyqIiMrmw4MYdG?usp=sharing

[DGM]

The file that's labeled true.tiff is actually a .jpg file.

I cannot reproduce any similar effects.

A = imread('true.jpeg');
h = imshow(A);
imwrite(h.CData,'ww.png')

B = imread('ww.png');

err = immse(A,B) % no change

err =

0

[Ali Almakhmari]

Sorry, apparently I was reading a jpeg file and saving it as a tiff file. However, you don't think thats whats causing the issue right? Cause I tried reading it as jpeg and saving it as jpeg and still I have the same issue.

[DGM]

Also:

A = imread('fake.tiff');
B = imread('true.jpeg');
err = immse(A,B)

err =

0

I do not see any difference in either of the two files, nor do I see any differences when saving any of them. I'm using R2019b/linux. They also look identical in other software.

What's strange is that the thumbnails on Google Drive look like two different images. It's not just that the color is different. They look like two different images.

Stack those two images atop each other and flip through them. I'd use a GIF, but color quantization would ruin the image. Pay attention to the position of the light (cloud?) features with respect to the darker background. I wouldn't expect any sort of color shift to cause displacement of object content.

... then again, it's a tiny, compressed thumbnail. I might just be seeing an artifact of the automatic thumbnailer's scaling/cropping efforts.

[Ali Almakhmari]

How do you not see any difference while saving them! Did you save them using the same piece of code that I mentioned above? I am starting to lose my mind. I am saving the "true" image and for same reason I get the "fake" one where clearly the colors are shifted just a bit.

[Walter Roberson]

When you read a jpeg and save it as jpeg, most of the time you end up saving it with different compression settings or compression algorithm details than were originally used.

Don't use jpeg for scientific work, unless the science involved is studying the jpeg algorithm.

[DGM]

I used the code as I posted above. I'm not sure what the difference is.

Difference in version/environment?

Difference between 'true.tiff' and 'true.jpeg'?

What happens if you try to save the file using imwrite() directly on pictureData instead of on h.Cdata?

[Ali Almakhmari]

I tried all your suggestions and they didn't help..except one. I think I will downgrade to MATLAB 2019b and see if it works. I really appreciate all your help. Let me know if you have any further suggestions.