Just as a remark to your point c: A watermark (visible or invisible) provides neither proof of ownership nor does it help crawling the web searching for the watermarked artifact.
Anyone can add a watermark to an image, regardless if they are the owner, creator, copyright holder or whatever.
Anyone can even add a getty watermark to an image. This is called a copy attack and it's the reason why the watermark is not very useful for identification by a crawler. A getty watermarked image could be a real getty image or any image someone inserted a getty watermark. In other words you could sort of DoS the getty crawler very easily.
Digimarc had an interesting research paper on their website about this topic but unfortunately it had been removed. In 2011 I wrote an email to Digimarc customer support asking if their newest watermarks are still vulnerable to a copy attack and the answer was: "There are hundreds of applications and specific use cases
for digital watermarking as I'm sure you found in your research. Some of our
published papers describe theories and some describe applications and solutions that
are available for license today as end-user products. "
A Getty crawler would have access to all the images that really are Getty images, making it fairly easy to identify a non-Getty images with fake Getty watermarks. Some sort of hashing function, perhaps, or something like tineye uses.
Since these new watermarks include a URL that identifies the Getty stock number of the photo, someone trying to fake that would end up putting a bogus stock number on the watermark, which makes it even easier to ignore fakes. The fake watermark would either use a number that doesn't correspond to a Getty picture at all, or a number that points at a different image.
I agree with you that the Getty crawler has access to all the Getty images and can identify non-Getty images. My point is that the watermark is not of much help for this task. On the one hand it is easily misled to find false positives on the other hand it will miss many images which have been altered strongly enough to destroy the watermark.
You are right with your objection that watermark contains the stock number. I don't see however how this can help a crawler to identify a fake stock number. To compare the (possible altered) image with the original we have to use an image hash and then we could have used the image hash in the first place.
Anyone can add a watermark to an image, regardless if they are the owner, creator, copyright holder or whatever.
Anyone can even add a getty watermark to an image. This is called a copy attack and it's the reason why the watermark is not very useful for identification by a crawler. A getty watermarked image could be a real getty image or any image someone inserted a getty watermark. In other words you could sort of DoS the getty crawler very easily.
Digimarc had an interesting research paper on their website about this topic but unfortunately it had been removed. In 2011 I wrote an email to Digimarc customer support asking if their newest watermarks are still vulnerable to a copy attack and the answer was: "There are hundreds of applications and specific use cases for digital watermarking as I'm sure you found in your research. Some of our published papers describe theories and some describe applications and solutions that are available for license today as end-user products. "
EDIT: Found the Digimarc paper http://www.google.com/url?sa=t&rct=j&q=&esrc=s&#....