I have two images. The first one is produced from the second one but I have to find how.

I have two images. The first one is produced from the second one and I have to find how this happens. Is there a way in MATLAB to do so? 1.png is produced somehow from 8B-1.6_30k_0d5_base_x=0mm_y=0mm_C001H001S0001000001.tif. I know the task is very difficult to handle but if someone gave me some clues I would be grateful.

Thanks a lot in advance

[Image Analyst]

8B-1.6_30k_0d5_base_x=0mm_y=0mm_C001H001S0001000001.tif.

1.PNG is below:

[Jan]

Sorry, I get different pictures to the "8B..." picture depending on the fact, if I open it from inside the unzip tool of if I unzip it at first. Please post both images using the "Image" button directly. This would increase the chance to find somebody, who had seen such images before, while downloading and expanding means to much work for most of the participants - as me, if I'm not able to display the file uniquely.

Answer 1

[Image Analyst]

I have no idea. Why don't you know? In general there is no way to know how an arbitrary image was produced from another arbitrary image. It looks like the second image has undergone some kind of high pass filter and has some saturated parts, but it doesn't look remotely like the first image. You might as well say you have a picture of your face and want to know what was done to it to turn it into a photo of the Grand Canyon.

[Efstathios Kontolatis]

I know it's very difficult. The second image is experimental X ray data. The first image is a process of that taken from another guy's algorithm which I don't have. I expect an answer from him but it won't be earlier than 2 months from now so I am trying to understand alone what has been done. However, believe me it's the same image.

[Image Analyst]

Maybe he used adapthisteq() but you'd have to do trial and error to determine the parameters he used. Or maybe he used conv2() with a high pass kernel, like [-1,-1,-1; -1,8,-1; -1,-1,-1]. Or maybe he used a bigger kernel - who knows. You might just have to wait until he gets done hiking the entire Appalachian trail or wherever he disappeared to. It looks like, from the almond shape, that perhaps the filtered image is a zoomed in and cropped and rotated version of the other image - perhaps, who knows? I can't even really locate the shape in the other image for certain.

Answer 2

[Jan]

It could be nearly everything: A principal component analysis, the factors of a fourier analysis, or even a SHA128 hash of the file name interpreted as grey levels. Most likely the file name is more useful than the image.

By the way, the image has very few information due to the pattern of stripes. It seems to be nonsense to store more than the first row.

My conclusion: It is impossible to find the answer of your question based on the images itself. There is an infinite number of mathematical operations to obtain the 2nd image from the first one, but very likely there is no way for the other way around.

So either wait for 2 month or give up.

[Image Analyst]

Another option: it could be a recovered watermark for all we know. Or a stegonographic image that was hidden/encrypted in the starting image and recovered.

[Efstathios Kontolatis]

@Jan I don't understand what you mean by 'By the way, the image has very few information due to the pattern of stripes. It seems to be nonsense to store more than the first row'. Can you please explain a little more?

Thanks you all for the answers