Due to the riches of the movie and music businesses, video and audio gain the most attention when it comes to online copyright violation. But unauthorised use of photographs and other artwork is also an issue, particularly for individuals with no corporate protection.

To address the problem of uncredited copying in the world of private and commercial blogs, a team at Tokyo's Waseda University has developed a system that can compare hundreds of thousands of images [Subscription Link] online and find duplicates.

Matching patterns

The first step looks for colour and angle patterns in the target photo among random samples of images on websites to find possible violations. After that, the system searches for five sections of the target photo among the candidates with a view to finding an actual match.

Demonstrations show that performing the analysis on samples of almost half a million images takes just five seconds, although the number of blogs already in existence suggests that the Waseda team will need to speed things up if it is to find a commercial application.