Queen Mary Vision Laboratory seminar
------------------------------------


Ontology-driven Inference and Fusion for Image and Video Analysis
By Christopher Town, University of Cambridge and AT&T Labs Research

Time: Wed the 28th Jan, 1-2PM
Venue: CS 338



Abstract:

My research aims to design and implement extensible computational models
for perceiving systems based on a knowledge-driven joint inference
approach. These models can integrate different sources of information both
horizontally (multi-modal and temporal fusion) and vertically (bottom-up,
top-down) by incorporating prior hierarchical knowledge expressed as an
extensible ontology.

Two implementations of this approach will be presented. The first consists
of a content based image retrieval system which allows users to search
image databases using an ontological query language. Queries are parsed
using a probabilistic grammar and Bayesian networks to map high level
concepts onto low level image descriptors, thereby bridging the "semantic
gap" between users and the retrieval system.
Secondly, I will present a sensor fusion problem in which computer vision
information obtained from calibrated cameras is integrated with location
events from a sentient computing system which uses ultrasound to track
people and devices in an office building. Fusion of the different sources
of information takes place at a high level using Bayesian networks to
model dependencies and reliabilities of the multi-modal variables. The
system maintains a world model which incorporates aspects of both the
static (e.g. positions of office walls and doors) and dynamic (e.g.
location and appearance of devices and people) environment. The world
model serves both as an ontology of prior information and as a source of
context which is shared between applications. It is shown that the fusion
of computer vision information derived using techniques such as image
segmentation, region and model based tracking, face detection, and image
classification enables the system to maintain a richer and more accurate
world model.



Speaker Biography:

Christopher Town is a third year PhD student in Computer Science at the
University of Cambridge, where he is being supervised by Prof John
Daugman. His work is sponsored by AT&T Labs Research through an Industrial
Fellowship from the Royal Commission for the Exhibition of 1851, and
through a research scholarship from Trinity College Cambridge.
Before starting his PhD, he spent 15 months working at AT&T Laboratories
in Cambridge and in the USA. Prior to that he received a Bachelors degree
with first class honours in Computer Science from the University of
Cambridge.