Browsing large image databases - Ronald Richter
- © Copyright??
- Given an image (middle), browsing allows interactively exploring the space of related images in a database of millions of images.
- © Ronald Richter
This thesis presents a software system that allows
users to browse large image databases that consist of millions of
images. Unlike existing search engines, which try to find images by
their textual annotations, this system only uses information contained
in the image itself. Furthermore, users are able to explore the
database by navigating into different directions of image
A subset of similar images are displayed on a touch-screen. By selecting one of them, a new subset of similar images is generated and displayed. Depending on the properties of the selected image (color, structure, etc.), the user moves forward to images with the desired characteristics. Because of the "looking around" approach, the user is able to locate himself in the database and thus can decide which path is leading to more meaningful results.
There are three key issues in this thesis: 1) Which information can be extracted from an image and can be used as a similarity measure. 2) How to query a database of millions of images in milliseconds, to keep the application interactive. 3) Which visualization and interaction methods can be used to increase usability.