Images have become one of the main sources of information to study
physical phenomena and to the design of intelligent systems. In their
most general form, these images can be multi-dimensional (e.g., CT and
MR images), multi-spectral (e.g., remote sensing images), or both
(e.g., digital video).
In any case, we are interested in image processing and analysis
techniques that enhance, encode, extract, represent, describe, and
classify the image content. Particularly, we have studied image
operators based on pixel connectivity for filtering, segmentation,
classification, compression, object representation and description. We
are also interested in special rendering techniques for
multidimensional image visualization and in the quantitative analysis
of biomedical structures for the purpose of diagnosis, treatment, and
research.
Content-Based Image Retrieval
Content-based image retrieval systems rely on image processing methods
that extract and represent image content in a compact way. We are
interested in shape, color, and texture descriptors for efficient
indexation and effective information retrieval from large image
databases. However, efficiency also depends on special data structures
for similarity-based search and low-level image features are usually
not enough to satisfy the users' expectations. Therefore, we are also
interested in the design of suitable data structures, which take into
account the user's preferences, and on relevance feedback techniques
to reduce the semantic gap.
Pattern Recognition and Machine Learning
Images, pixels, or image objects can be considered samples of a
dataset. We are interested in clustering, pattern classification, and
machine learning techniques for general data analysis and for image
processing applications, such as restoration, compression, and
retrieval. Particularly, we have investigated methods that exploit a
connectivity function between samples in the feature space.
Digital Video and Biomedical Imaging Applications
A digital video is a sequence of color images along the time. This
makes of it a 3-dimensional and 3-spectral data set. Our interest is
on graph-based video processing techniques for compression, retrieval,
segmentation, and object tracking.
In medical imaging, sequences of digital images form 3D/4D data sets
containing object systems under study. The visualization and analysis
of such objects (e.g., organs, tumors) usually require filtering,
interpolation, alignment, registration, segmentation, representation,
description, and classification of multidimensional data. We are
interested in such techniques applied to magnetic resonance images of
the human brain for the study of epilepsy and other degenerative
diseases. We are also interested in the automation of the diagnosis of
intestinal parasites by the analysis of microscopy images.
Last time this page was updated and we remembered to update this line:Aug 2nd, 2004