Searching for images is about to become a whole lot like, well, searching for anything at Google (Nasdaq: GOOG).
Company researchers presented a method for improving image search results to attendees of the International World Wide Web Conference in Beijing. Google calls the process VisualRank, after the company's signature PageRank technology.
VisualRank, The New York Times reports, is an algorithm that blends image-recognition software methods with Google's own process for weighting and ranking images that appear similar.
Image search technology isn't new, of course. Social network Flickr, now a part of Yahoo! (Nasdaq: YHOO), uses tagging to collect photos into relevant bunches. And both Microsoft (Nasdaq: MSFT), at Live.com, and Google already have engines tuned to search for images. The problem is one of irrelevant results.
For example, searching for "Iron Man" and "movie" at Google image search brings you both movie posters for the forthcoming film as well as old comic book covers featuring the iron avenger.
But you knew that. You've known for years of Google's quest to improve its search capability and the quality of its services. You know, too, that Baby Breakers are emerging to challenge Google in search. Why should we believe VisualRank is the breakthrough that DoubleGoo's researchers believe it is?
Perhaps we shouldn't. Search isn't perfect, after all, and likely won't be for some time.
Then again, that's a pedantic view. All progress takes time. What we know from history is that, in computing, features improve as processing power grows.
Google, arguably, has the world's largest networked computer and thereby an ability to crunch a lot more data than most. Now, a portion of that horsepower is going to be applied to a more advanced image search algorithm.
Image search ads? They're about to be a click away, Fool.
Get your clicks with related Foolishness:
Get the best of the Fool delivered to your inbox every Friday
Terms & Conditions
RSS Headlines
Fool UK