Computer system searches for images based on hand-drawn sketches
While searching the web is definitely more efficient than it once was thanks to search engines like Google and Bing, sometimes searching for just the right image is a bit more difficult.
That’s why computer scientists from the University of Basel developed a new, open-source method for conducting image and video database searches based on hand-drawn sketches.
To operate the system, a user can draw a sketch on a tablet or piece of interactive paper and the system will search for a matching image in the database.
The system, called vitrivr delivers the images and video clips that most resemble the hand-drawn sketch. For videos, the user can even specify on the sketch in which direction an object is moving in the searched sequence to help produce more targeted results.
When designing the system, the researchers deliberately set a very broad similarity concept and adapted it to different types of sketch (for example, similar colors, shapes or directions of movement).
Individual searches can be filtered by a range of other query types, such as search terms, examples of images and videos, or combinations of these. The team considers the new system’s scalability one of the most important features because it can be used even with very large multimedia collections.
The vitrivr system is entirely open source and freely available to international researchers. It is currently being used for a variety of purposes, such as discerning patterns of movement in sports videos and searching collections of digital watermarks.
Researchers around the world are working to further develop the system, with the support of programs like Google Summer of Code.
Comments are closed, but trackbacks and pingbacks are open.