| Author | Thread |
|
|
02/03/2006 11:58:20 AM · #1 |
Binghamton University (SUNY Binghamton) is working on a process to link digital images with the cameras that made them. You can read about this interesting research here: //inside.binghamton.edu/news/newspage.cgi?issue=2006feb02&id=3
|
|
|
|
02/03/2006 12:19:58 PM · #2 |
Interesting... this group has patented a way to reliably identify the camera used to take a picture. You have to have the suspect camera and compare images taken with it, but it will work. It can also be used to detect image forgeries.
It was originally devised to catch child pornographers but could be used effectively in a lot of other useful ways.
|
|
|
|
02/03/2006 12:30:01 PM · #3 |
| i wonder what the effects of interchangeable lens's on dslr's would be (if indeed any at all)? |
|
|
|
02/03/2006 12:35:03 PM · #4 |
| Very interesting, but I wonder if this fixed-pattern-noise is sufficiently stable to guarantee recognition of the same camera across months/years. Also, noise is always affected by temperature. My company is doing similar research with infrared cameras (thermal, not NIR), where the noise changes in a matter of hours. Note that they've only tested this on nine cameras so far, sounds like a good idea that needs a lot more testing. |
|
|
|
02/03/2006 12:42:52 PM · #5 |
Interesting idea but like some of the previous comments, wonder how consistent the noise patterns are over time on the same camera.
I also wonder how digital manipulation would affect it - think of RAW conversion or noise reduction or anything really in a lossy format.
It bothers me a bit that they compare to finger prints cause it might not be that conclusive (it may turn out that way but thats a leap without more verification) but thats probably the filler stuff for the reporting.
Forgot to add: It might be a big help in image theft because it might give you a way to prove the image was from your camera or matched your other photos. I know there are currently some digital "fingerprints" that can be imbedded in photos but as far as I know they are not common.
Message edited by author 2006-02-03 12:45:32. |
|
|
|
02/03/2006 12:43:39 PM · #6 |
nt
Message edited by author 2006-02-03 12:52:01.
|
|
|
|
02/03/2006 12:49:38 PM · #7 |
It doesn't sound like this would work after a liberal application of edits has been performed on an image.
Such as NeatImage, USM and other actions/filters that will actually modify the pixels that are present in the images being looked at.
Am I right or am I wrong?
It's a great concept, I just don't see how well this could work out. |
|
|
|
02/03/2006 12:52:26 PM · #8 |
That's actually something totally different, called check-sum verification which has been around for decades. That's to verify that two files are identical, wheras that Birmingham U work is about verifying that two totally different images came from the same camera. |
|
|
|
02/03/2006 12:56:16 PM · #9 |
Originally posted by magnus:
That's actually something totally different, called check-sum verification which has been around for decades. That's to verify that two files are identical, wheras that Birmingham U work is about verifying that two totally different images came from the same camera. |
My mind was triggered by the word 'forensics' and I was looking at it from the point of the investigator who has to prove that his images are original. Only after that I began reading the article and read their intentions. My mistake.
And I think that when criminals find out about this noise-fingerprinting technology they'll simply develop a countertechnology.
I also wonder how you are going to prove this link to a camera when someone uses chroma NR, pattern NR and after that resizes their picture to 50% to publish it on the web or to put it on a CD. Only someone stupid would get caught.
Message edited by author 2006-02-03 13:06:35.
|
|
|
|
02/03/2006 01:00:47 PM · #10 |
Originally posted by magnus:
... wheras that Birmingham U work ... |
Ahem, that's Binghamton (a place in New York), not the Alabama folks.
=) |
|
|
|
02/03/2006 06:11:16 PM · #11 |
Originally posted by KaDi:
Originally posted by magnus:
... wheras that Birmingham U work ... |
Ahem, that's Binghamton (a place in New York), not the Alabama folks.
=) |
Sorry! I actually thought it was Birmingham, UK! Sure, I can read...
As for surviving edits, noise reduction in particular, that's a good question. But my guess is that just like a real fingerprint, you can still do okay with just partial matches. In other words, if the fixed-pattern noise matches perfectly for even just a small uniform region somewhere in the image, then you can demonstrate that it would be statistically highly improbable that the image came from another camera. |
|
Home -
Challenges -
Community -
League -
Photos -
Cameras -
Lenses -
Learn -
Help -
Terms of Use -
Privacy -
Top ^
DPChallenge, and website content and design, Copyright © 2001-2025 Challenging Technologies, LLC.
All digital photo copyrights belong to the photographers and may not be used without permission.
Current Server Time: 11/27/2025 07:12:33 AM EST.
