Copyright protection for JPEG images

Automated reporting of copyright violation (just an idea)

[Scooter24]

I have this idea and am interested in your opinion:

Imagine JPEG images contain meta information about the owner and his site. Could be a watermark or something else. Must be speedily extractable by a web server and can be removed by the owner (in case the owner sells the image).

Imagine also that a new feature is added to the Apache web server software. Let's call it 'unauthorised reproduction notification'. It works as follows´:

- when the server receives a GET request for a JPEG image from a browser it checks if the image belongs to the site where it is hosted (by comparing the owner meta data in the JPEG file with the web address of the site where the image is hosted)
- if there is an unauthorised reproduction, the server automatically notifies the legitimate owner, for instance by email

Of course this copyright protection could be removed by a knowledgeable person, but it would work in a % of situations.

New releases of the Apache web server (and possibly other server software as well) would however have to incorporate this feature.

[aus_dave]

Interesting idea Scooter ;).

Not sure if you know about Digimarc technology but it is a stand alone product that does a similar thing to what you suggest, but not on a server. The artist/photographer can put a digital signature 'inside' the JPEG which stays with it when downloaded or emailed. You get an individual key that you can use to verify an original image is one of yours when contronted by a copyright infringement. Photoshop can display Digimarc info contained in an image.

If you subscribe to the premium version I think it can also track your images across the web - amazing stuff. At the moment I am using a visible watermark to protect my images but am looking seriously at Digimarc for the future.

[hakre]

there is a method to add and remove meta data to jpeg images. this is build in. most digital cameras do so. you can use irfanview for example to view that data (and delete it).

[Scooter24]

Digimarc unfortunately is quite expensive and it doesn't alert you automatically when a copyrighted image is republished in the Internet.
Besides what does Digimarc do - scan on daily basis all 2+ billion pages in the Internet? Seems impractical.

[Tapolyai]

Take the data area of the image and combine it with a private key to generate a public key. Place the public key in the meta tag area of the image. Image can be only displayed if the public key decodes with provided key from author, otherwise displaye the image and overlay it with "unauthorized" or something. Possibly encrypt data area of the image with the private key, not allowing display at all.

Does not require hardware, only software at the server. Requires publisher to contact author. Possibly build in automatic contact of author to make micro-payment every time image is displayed.

Standard in other security areas. Public/Private key method.

[hakre]

so that's the technical basis. whatif noone wants to setup such a server? and: who defines who owns the copyright. and today: to whom should we pay the fees of using another patent? it's totally impracticable.

the other way round is under strong development: image recorgnition. when google will have the option in image search: find similar images (because of their look), then your own law unit can search the web for your works and derivations of it. i think, this is what would come up earlier then a special webserver edition.

[Scooter24]

image recorgnition. when google will have the option in image search

Interesting. When will Google have this feature?

[hakre]

hi scooter,

i think that maybe pre-beta at google at all. some computer scientists all over the world try to do that thing. if i remember correctly last year they got a little breakthrough and have shown some betaapplications. worked quite well.