A photo says many stories, but unlike text, the context of a
photo is hard to search for unless explicitly "translated" by a human
being. The photo web of today is like the text web before Google came into the
picture. Sweden-based Polar Rose will make photos searchable by analyzing their
content and recognizing the people in them. The company intends to offer free
software to make photos of familiar faces searchable on both personal computers
and on the web by analyzing the contents of pictures with pattern recognition
technology to locate specific faces within them.
It would allow users to annotate photos with descriptive
details, using the web to improve what can be done with computational searching
alone on sites like Google or Yahoo. Polar Rose, which takes its name from a
flower-shaped mathematical curve used to plot two-dimensional coordinates, will
help consumers label any photo and thereby make it easier to search for related
photos of same or similar-looking people. In simple terminology, Polar Rose will
help the computer user to sort through and group his personal photos face by
face. More broadly, the software can also search for similar-looking photos
across the web. The software analyzes digital photos to locate faces then
converts the data from two-dimensional (2D) images into 3D models. These
skeletal models can be rendered into what scientists call "faceprints."