I think it's one of the most satisfying feelings as a photographer to find one of your images in use somewhere. I've come across mine in magazines, on BBC news and even on altavista many years back!
TinEye is a visual search engine which allows you to search for your images by submitting an image for it to analyse, it returns a series of hits which are pages where it thinks it has found a match for your images. It works by comparing the pixels and shapes in the images not by searching for meta tags or matching file names.
Currently tin eye has indexed just under 500 million images, which is only a tiny fraction of all the images online, As an example if I search on tin eye for one of my images I might get 2 results, if I search for the file name in google I get dozens more results from users who have used the image and left the original file name intact, there are likely many many more than this in use where the file name has been changed. If tineye covered as many sites as google then tineye would return more results, as it is able to locate images which have been cropped, photoshopped or had text added to them.
This not a new idea, I've seen companies offering the service before, but for a fee, many of them do it to capture unauthorised use of corporate logos. At the moment TinEye is in beta development and there is no charge for it, you need to request a membership (it took a day when I requested membership).
Success and Failure
I tested tin eye with a selection of my photos which have been around on the internet for a while offered as free samples, so before I started I had a pretty good idea that the site would find some hits. And indeed it did. I found my images scattered all over the place! (and I suspect they were not microstock purchases or properly credited as they should have been - but that's what you get offering some free samples...)
I tried testing with an image of a padlock:
Initially, I submitted a thumbnail of the image and tineye returned one match, which was indeed a copy of my image. I then tried submitting a high resolution version of the same image, this time tineye returned two matches, including this one:
On close inspection I could see that the returned image was indeed a cropped, coloured and modified version of my image. Really quite impressive stuff!
I did have less success with an image of the Eiffel tower, probably because it is so well photographed, I found several matches to my original image:
For this photo search TinEye returned 12 results, none of them were my images, but I'm not complaining as they could be modified versions of my image, and they are all images of the Eiffel tower taken from a similar angle with its distinctive shape. As I understand it if any of them did more closely match my image then they would appear at the top of the search results. The more different the images are the lower down the results they display, here are a sample of the results returned:
Probably the best feature tineye offer is their firefox plugin, it makes it as simple as navigating to any page (e.g. one with your images on them) right clicking and selecting Search Image on TinEye. This makes checking your images much easier if you have a gallery of them online at a reasonable resolution that tineye can access. You could also use the watermarked versions from one of the microstock sites, BUT results would be much less reliable.
I had great fun matching my images up in all sorts of places. Due to the limited size of the current image index it's not really possible to use tineye to perform any sort of license enforcement checking or measure how frequently used your images are, I had plenty of images that returned no results which I know for certain are on other websites. I'm looking forward to the search index covering a larger proportion of the images online (no mean feat with all those billions(?) of images!). I'm also looking forward to any premium features that tineye might have planned such as automatically and regularly checking for unauthorised copying of your images perhaps from a selection of images that you have uploaded to their server.
An alternative commercial service for 'visual content owners' (their words not mine) is picscout. Their (paid) service is more directed towards rights managed image use and probably not of too much interest to microstock sellers.
Update as of JUL 2012 tin eye has indexed 2.1 billion images and registration is no longer required to access the service.