Image Specialists - please help in image analysis ideas

On some greyscale images released to the public by NASA, there appears to be an 'overlay' or mask which is placed on top of the base image.

Is there any way of detecting the 'mask' that is added to or overlaid on each compressed photo by analysing pixels from many (100's) photos?

I realise that each greyscale image is made up of individual pixel values which in the RGB colour method, show the intensity of grey which ranges from pure black to pure white.

So assumptions (which I am not sure can be assumed, but anyhow...)
1) NASA add the 'overlay' AFTER they have compressed the images into jpg format.
2) the 'overlay' is the same for each image and is NOT a function of the image itself.
3) the bit pattern 'overlay' will not be subtracted/added to pixels which are pure black or white of near the ends of the greyscale.

then by subtracting one image from the next and ignoring the top 5%(?) in the histogram of the greyscale, we should be able to arrive at a base value for each pixel position.

If this base value obtained this way is added back to each picture, making that pixel lighter, would the image become clearer/sharper/better?

Can we assume that 1, 2 & 3 are true - possibly not?

Any ideas or comments whether this would be possible please?
Thanks
Q

Originally posted by qmantoo
On some greyscale images released to the public by NASA, there appears to be an 'overlay' or mask which is placed on top of the base image.

Let's start with this first.

What causes you to say this?

reply to post by qmantoo

I'll share with you what I was told by my Photoshop/Image Editing instructors when I asked almost that precise question a few weeks ago. They know what I do in my off time online and understood why I was asking....they also weren't the least bit helpful.

First.. Every image of any meaning or importance of ANY kind is almost 100% certain to have been photoshop'ed. That's magazines, newspapers (especially them), pictorials and television. Everything. They'd need to make a major decision NOT to these days. It isn't to do things to millions of photos we all see that changes them in any material way. 99% aren't being altered for dishonesty. In fact, much of what I'm learning in Photo preparation and editing is SO subtle and so hard to see, even when I'm the one who did it, it's easy to see it's strictly for quality and presentation.

Second though....... No. You can't tell. I didn't just ask one, I asked both I have courses with this year and then asked the Dept Head. All came back the same. If someone is good and takes their time to do it correctly, there is ALMOST no way to tell an image isn't real and originally as you see it. If they take that extra time and really go the distance, there IS NO way to tell at all.

The exception to telling in the "almost" is zoomed 1 step above pixelation and the grid appearing on PS or other high end programs. I.E....Your SO zoomed, nothing is image anymore. It's thousands of shades of color in blobs. If someone didn't do a professional job...those blobs won't transition and merge from one original color to another. There will be a few pixels in line to the edges that are just a BIT more abrupt for transition that everything else in the image......and that's what it takes IF they were sloppy. So the college folks tell me, anyway.

Originally posted by Wrabbit2000
First.. Every image of any meaning or importance of ANY kind is almost 100% certain to have been photoshop'ed. That's magazines, newspapers (especially them), pictorials and television. Everything.

That seems off-topic if we are talking about raw NASA images.

Yes, NASA does photoshop some images for publicity purposes, but they don't claim they are raw images.

If you look for raw images from NASA, they are not supposed to be photoshopped.

reply to post by Arbitrageur

Sorry if the anyone figures I strayed... He seemed to be asking if it was possible to detect the deliberate modification of a real image, digitally. I.E.....can the mask be detected in a provable and definitive way?

I was trying to be helpful.... I rather thought the time they all took explaining to me how it actually works bore on the question? err...

Originally posted by qmantoo
On some greyscale images released to the public by NASA . . .

Does NASA keep the public guessing whether the image is raw or not?

What is NASA's convention for identifying images?

Ok, well it does not have to be NASA photos as I am asking a general question. So please dont focus on NASA, I just happened to mention them as an example.

What I am trying to find out is whether there is any way to determine what 'mask' or 'overlay' has been used on a greyscale jpg image. I am just using the NASA images as examples and where there are many available.

I see that some images have been compressed and the patterns on the image look the same. Now... in theory in my mind, I think that if I can get a pure sky in two similar photos, I should be able to isolate the pattern in both (which looks the same to me) and then subtract it or add it to the images to get a better picture.

Is this possible or am I totally barking up the wrong tree ?

Originally posted by qmantoo
I see that some images have been compressed and the patterns on the image look the same. Now... in theory in my mind, I think that if I can get a pure sky in two similar photos, I should be able to isolate the pattern in both (which looks the same to me) and then subtract it or add it to the images to get a better picture.

Is this possible or am I totally barking up the wrong tree ?

It would be a lot easier to answer if you gave us an example.

There is a thread on ATS about on overlay of sorts on NASA photos, put there by accident. It was some kind of dirt and debris in the photographic mechanism. An ATS member did a brilliant job of making a GIF to show how this overlaid different images taken by the NASA camera.

www.abovetopsecret.com...

Originally posted by Phage
[atsimg]http://files.abovetopsecret.com/images/member/d84fa289706e.gif[/atsimg]

This is the closest thing I can think of to what you're talking about. But if you have an example of something different you want to analyze I'd take a look at the photos.

Originally posted by Mary Rose
Does NASA keep the public guessing whether the image is raw or not?

What is NASA's convention for identifying images?

That question could be answered with a little research. For example:

Raw Images-See the Latest from Mars!

Raw image identification doesn't get much clearer than calling them "raw images", right?

But the rule of thumb is, if you just see some random NASA image, especially if it's in a news article, don't assume it's raw. Like any photo for publication, it may be enhanced for best appearance in the publication. It could even be a composite of different raw images. But in every case I've seen so far, if you want to track down the raw image(s) you can.

In the case of old photos on film though this is extremely difficult, as the film is fragile. It's been scanned but there could be scanning defects not in the original raw film image, like if there was one bad pixel in the scanner for example, this could create the appearance of some kind of mask or overlay in multiple images. But a researcher can probably get access to the film if there's a good enough reason. It's a lot simpler with the latest digital photos, as there's no film or scanning involved.

reply to post by Arbitrageur


My point was: What is NASA's convention?

They have their own website. For their own communication with the public, without a middle man intervening, what is their policy when posting images? Do they feel under any obligation to be specific about whether or not the photo is a raw image?

I see that some images have been compressed and the patterns on the image look the same. Now... in theory in my mind, I think that if I can get a pure sky in two similar photos, I should be able to isolate the pattern in both (which looks the same to me) and then subtract it or add it to the images to get a better picture.

You're not entirely wrong, but unless you have perfect or at least incredibly similar pixel values in both images or they're of the same location it's not going to provide amazing amounts of useful data likely. You would need to somehow be able to tell the difference from a 245 pixel that's part of the pattern and a 253 pixel that is difference in the sky. You would need to do some analysis and get lucky with the similarities.

I get the impression the pattern is hard to see due to the grey scale? You could always just go for the classic histogram equalization and see if you can't just bring up the pattern to a visible state for comparison. Isolate it, and then treat it like a basic ghetto matte pull. Though with the basic work alone you should be able to compare them.

As has been said though, you really need to show an example for these types of things since it will strongly change ones approach in analysis/processing.

Originally posted by Mary Rose
reply to post by Arbitrageur

 

My point was: What is NASA's convention?

They have their own website. For their own communication with the public, without a middle man intervening, what is their policy when posting images? Do they feel under any obligation to be specific about whether or not the photo is a raw image?

I gave you a direct link to where they post raw images and call them raw images.

If you see the images elsewhere, then my opinion is they are beyond the control of NASA, so they would probably only assure us the raw images are the ones we download from their raw images link. I'm not aware of their exact policy but you could certainly ask them this question or search to see if they've already answered that in a faq somewhere. It's not a question I'm inclined to ask as it's already pretty clear to me...if I want raw images, download them directly from the NASA raw image link, and if the image appears anywhere else, especially in the media, the status is uncertain. If some website was abusing NASA images, such as representing an image as raw if it's not, and you complained about that they might take action. But my guess is they would tell you what I did, that if you want raw NASA images, get them from NASA's source labeled as such, and not from somewhere else...it's not that hard.

A common example that comes to mind are images on Google moon...Google uses stitching and compression algorithms which distort the images, and while they may show NASA as the source, which is true, that's before the algorithms are applied, so what you get on Google moon is a lot of distorted images, definitely not raw. And NASA probably doesn't feel any obligation to explain all that since anybody can research it. (if they are a good researcher).

reply to post by qmantoo


What do you mean by 'mask' or 'overlay'? What exactly is being masked/overlaid? I am familiar with these terms, but your context is not complete enough to make any meaningful commentary.

There are many image lossy compressing algorithms, each producing aliasing artefacts when pushed beyond their best case. The aliasing is source-dependent, the similar "patterns" you are seeing are most likely from similarly clustered pixels, of course what makes them similar is algorithm-dependent.

Originally posted by Arbitrageur
y point was: What is NASA's convention?

They have their own website. For their own communication with the public, without a middle man intervening, what is their policy when posting images? Do they feel under any obligation to be specific about whether or not the photo is a raw image?

I gave you a direct link to where they post raw images and call them raw images.

If you see the images elsewhere, then my opinion is they are beyond the control of NASA, so they would probably only assure us the raw images are the ones we download from their raw images link. I'm not aware of their exact policy but you could certainly ask them this question or search to see if they've already answered that in a faq somewhere. It's not a question I'm inclined to ask as it's already pretty clear to me...if I want raw images, download them directly from the NASA raw image link, and if the image appears anywhere else, especially in the media, the status is uncertain. If some website was abusing NASA images, such as representing an image as raw if it's not, and you complained about that they might take action. But my guess is they would tell you what I did, that if you want raw NASA images, get them from NASA's source labeled as such, and not from somewhere else...it's not that hard.

A common example that comes to mind are images on Google moon...Google uses stitching and compression algorithms which distort the images, and while they may show NASA as the source, which is true, that's before the algorithms are applied, so what you get on Google moon is a lot of distorted images, definitely not raw. And NASA probably doesn't feel any obligation to explain all that since anybody can research it. (if they are a good researcher).

reply to post by Mary Rose


My question to you is only about NASA.

A link where they call them raw images does not answer my question, exactly.

Perhaps you don't know.

Nevermind.

Originally posted by Mary Rose
reply to post by Mary Rose

 

My question to you is only about NASA.

A link where they call them raw images does not answer my question, exactly.

Perhaps you don't know.

Nevermind.

If you clicked the link, they are clearly labelled as raw images. What more do you need to know?

Originally posted by Mary Rose
reply to post by Mary Rose

 

My question to you is only about NASA.

A link where they call them raw images does not answer my question, exactly.

Perhaps you don't know.

I admitted I don't know their policy. But I do know what I've seen in practice. Even on NASA's website they may take black and white raw images taken at different wavelengths, and try to make a color image out of it, even though it's not true color. Here's an example:

Herschel Crater in 3-D
In that case they explain the origin of the image but I'm aware of no policy at NASA that requires them to explain the origin of each image they post in that much detail. I think they have other color composite images posted where they don't explain as clearly, that all the original raw images are in black and white.

Mary Rose - tthere is a good article here from Emily at Planetry org which explains what "raw" actually means and is NOT the same as "best quality" which eventually goes into the PDS for archiving. Raw, as I understand it is the un-stretched and "quick and dirty" version of what goes into the PDS.

planetary.org...

To everyone, I will try and find a good example and give a link. Thanks for the comments and suggestions so far.

sorry I have left it too long to edit my last post.

I can give you an example of a NASA Mars Rover image if you like.
marsrover.nasa.gov...

If you zoom in to the rocks on the right, the black areas have a kind of overlay around the bottom edges and overlapping the black a fair bit. While there, you can see this 'overlay' extending to the rest of the ground. This is not a regular compression instance as it would either compress the black area or it would not compress any of the black area.

Maybe it would be possible to find areas which are overlaid over black(a known 'colour') and to determine the pattern from that?

reply to post by qmantoo


Where are your image specialists?

yes mary, a good question. This topic is possibly getting hot and slightly controversial and being in the science forum may not attract those willing to discuss or debate that side of things. Dunno though, they may turn up...:-)