Bridging the gap between human
and computer abilities to interpret image data and making image understanding
technologies an integral part of personal computing.
With the widespread and cheap availability of imaging
devices (scanners, digital cameras, camera-equipped PDAs and
cellular phones), and the rapid decrease in storage costs,
individuals and organizations are acquiring and collecting vast
amounts of image data. Yet, operating system and application
software has not caught up with this development; imaging is still
the domain of special-purpose applications.
The long-term goal of this project is to bridge the gap that currently
exists between human and computer abilities to interpret image data and
to make image understanding technologies an integral part of personal
The initial goal of this work is to create a flexible toolbox for
document layout analysis, optical character recognition,
handwriting recognition, image matching, and content-based
retrieval based on a class of geometric algorithms and
statistical methods being developed in our group. This toolbox
will also be demonstrated using a number of sample applications
in document and image management, personal digital libraries, and
context-aware information retrieval.