JOSA A is planning a special issue on Bayesian and
statistical approaches to vision to be published in summer, 2003. I
have enclosed the announcement below and would like to encourage
interested researchers to consider submitting a paper for the issue.
The deadline for submission is Oct. 1, 2002. If you have any
questions about submissions, you can contact one of the guest editors
listed at the bottom of the announcement.

David Knill, Bill Geisler and Bill Freeman (guest editors)

----------------------------------------------------------------------------
--------

JOSA announcement:

                The editors of JOSA A are soliciting papers for a
special issue on Bayesian and statistical approaches to
vision. The special issue will span topics in human visual
perception, computer vision and neural coding of visual information.
 The past decade has seen an explosion
of interest in the application of statistical techniques to modeling
vision in both biological and artificial systems. Vision is
fundamentally a problem of making inferences about the world from
uncertain information. How sensory noise and statistical regularities in
the environment structure problems of visual coding is central
to a computational understanding of vision. The mathematics of
statistics, stochastic processes and Bayesian inference provide the natural
framework for understanding these aspects of visual processing.
They have also provided a fertile framework for linking computational
theories of vision problems to models of information processing in
biological systems.
 In the domain of computer vision,
recent advances in algorithms for performing optimal statistical
inference have opened the door to a broader array of implementations of
working systems built on rigorous statistical characterizations
of visual problems. These have led to more robust and effective
algorithms for problems ranging from 3D estimation to object
recognition. In biological vision, statistical signal processing has
provided the framework for rigorous models relating neural coding to the
statistical structure of natural images. Researchers have extended the
application of ideal observer models to higher level visual problems
such as perceptual organization, depth perception and
object recognition. Bayesian models of perceptual performance have
also begun to emerge to explain a wide assortment of
psychophysical phenomena in higher level vision.
 Suggested topics for submission
include, but are not limited to
Natural image statistics and computer vision
Natural image statistics and neural coding
Bayesian approaches in computer vision
Applications of ideal observer models
Statistical / Bayesian models of perceptual performance
Applications of decision theory to perception
Signal detection models of recognition and attention
Statistical learning in machine and biological vision

Feature Editors
                                  David Knill
                                  University of Rochester
                                  Rochester, New York
                                  knill@cvs.rochester.edu

                                  William T. Freeman
                                  Massachusetts Inst. Of Technology
                                  Cambridge, Massachusetts
                                  wtf@mit.edu

                                  Wilson S. Geisler
                                  University of Texas
                                  Austin, Texas
                                  geisler@phy.utexas.edu