Vision of Google image search future were uncovered recently by two Google scientists Yushi Jing  and Shumeet Baluja. Google previous technology improving image search and using human efforts Google’s Image Labeler though innovative to some point is no real solution to the task of image search quality . The next generation Google image search technology goes beyond indexing text associated with an image.

Now Google is going to use computer analyze in photos and new Image PageRank algorithm to relate it in a ranked way with keyword queries. Google is studying visual characteristics of popular images to assign them the highest Image PageRank and then determining other images rank based on similarities between images. The automatic analyze of image content will reveal a new dimension of content still not visible on the Web.

Google scientists assure that Google is now ready to apply high-tech image recognition software to digging into the images themselves. Images related to people and celebrities may rely on face recognition/similarity, images related products may use local descriptors, other images such as landscapes, may rely on color information, etc. The results of experiment based on 2000 images of the most popular products queries «show significant improvement, in terms of user satisfaction and relevancy, in comparison to the most recent Google Image Search results”.