When Google introduced its PageRank algorithm long ago it allowed web searchers to have a metric they could look at and easily determine the authority of a webpage. Google researchers are now saying they have developed technology to do for images what PageRank did for web pages.

The New York Times reports that a pair of Google scientists presented a paper called “PageRank for Product Image Search” at the International World Wide Web Conference in Beijing. The software technology is being called VisualRank and is at its core an algorithm that blends techniques for recognizing images and technology for weighting images and ranking them based on what looks the most similar.

Google already has an image search engine that is widely held to be one of the largest image databases online. The current image database pulls images based on clues from text associated with each image. This for instance is why you might get an image of President George W. Bush if you did an image search for Republican.

What the paper the Google researchers presented proposes is a method to actually rank images based on things in the image. Technology has been in place to recognize faces in images for a while, but identifying other things by computer in an image that humans can identify at a glance like a car or mountain has lagged.

Google researchers Shumeet Baluja and Yushi Jing told the New York Times, “We wanted to incorporate all of the stuff that is happening in computer vision and put it in a Web framework.”

The Google image database is too vast to apply the visual search algorithm to all the images indexed. What the researchers and a staff of 150 Google employees created was a scoring system to show image relevance concentrating on the 2000 most popular product queries including things like the iPod, Xbox, and Zune.

Google staff sorted the top 10 images from its normal ranking system and the Google Image Search results to get the image relevance and create the required scoring system for image relevance.