Google unveiled a prototype algorithm at a conference in Beijing last week that will add precision to the search engine’s image-search technology, The New York Times says.
Two Google researchers presented a paper describing the prototype, which is called VisualRank. It uses image-recognition technology to help rank the relevance of images found in a search.
Currently, Google Image Search results are ranked using the text acircular the image on the page. The new method will use the visual characteristics of the image itself, and rank search results by comparing similarities among them. —Hurley Goodall