Statistical and Mathematical Theories for Computer Vision
 
Session co-organizers:
Qiang Ji
Assistant Professor
Dept. of Computer Science
UNR
qiangji@cs.unr.edu
 
Robert M. Haralick
Professor, Dept. of Electrical Engineering
University of Washington
haralick@ee.washington.edu

An image understanding (IU) algorithm infers the state of the
physical world from the inherently noisy and ambiguous images of
the world. To operate accurately and robustly, an IU algorithm
must be able to model, account for, and propagate uncertainties in a
systematic and consistent manner while using the most complete prior
information that is available. Even in the situation of conflicting
information, an IU algorithm should infer a consistent and correct 
interpretation of a 3D scene.  Moreover, the general  ill-posed nature 
of interpreting the scene from image data further complicates the 
inference process.
 
These difficulties provide an ideal environment for the application of the
established theories in statistics and mathematics.  In recent years, 
we have witnessed a significant increase in the use of statistical and 
mathematical theories to pose and solve computer vision problems
as optimization problems. The objectives of this session are : 
1) to bring together researchers in computer vision, photogrammetry, 
statistics, mathematics, and optimization theory to showcase the latest 
developments in applying statistical, mathematical and optimization theories 
to a wide range of computer vision problems; 2) to promote the rigorous 
statistical and mathematical treatment of computer visions problems and 
to emphasize the role of mathematics, statistics, and optimization
theory as a rigorous basis for computer vision; and 3) to promote 
interaction and collaboration among researchers working in the fields of 
computer vision, statistics, mathematics and optimization theory.
 
  The scientific programme of the session will include  invited papers 
and contributed papers.  Original papers are solicited on topics 
including, but are not limited to: 
 
 Bayesian statistics 
 Statistical evaluation and validation of computer vision algorithms
 Imaging error modeling, analysis, and propagation
 Statistical object recognition
 Statistical feature extraction
 Statistical approaches for camera calibration, pose estimation,
    motion estimation, feature matching, and 3D reconstruction.
 Statistical image pattern recognition
 Robust computer vision
 Non-linear least squares
 Optimization theory
 Stochastic modeling and optimization
 Statistical approaches for sensory fusion.