VisionScienceList: Sussex University postdoctoral position

From: Roland Baddeley (rjb@biols.susx.ac.uk)
Date: Tue Jan 04 2000 - 10:05:57 PST


Applications are invited to apply to work on an EPSRC funded project
``Early visual processing and the effects of variable illumination'',
studying the implication of the natural illumination variation on
human low level vision using computer modeling and Bayesian statistics
methods and the collection of natural image statistics.

This post-doctoral position is available for 3 years starting
date Monday March 20th 2000. Final application date Monday January
28, 2000.

The position will involve working with Dr Roland Baddeley and Dr
Daniel Osorio at the School of Biological Sciences in the laboratory
of Experimental Psychology, and Neuroscience, Sussex University. The
applicants will be expected to have experience in programming (they
will be using MATLAB but this can be learned). Experience in any
of the following would be useful: colour or lightness constancy,
natural image statistics, image processing, visual psychophysics,
computational modeling of visual processing, neural network modeling.

This project will involve two stages. Firstly a data base of natural
images will be created. By covering all possible surfaces with a
coating of known reflectance, and then placing a card of known
reflectance and orientation at all locations within a scene, for each
image we will work out the relative contribution of the illumination
it is possible to decompose the final image into the contributions
from illumination, surface orientation, and surface reflectance.

This data base will be used to explore the hypothesis that much of
early vision is concerned with extracting the contributions of the
underlying reflectance and surface orientation from the original
intensity image- in other words that illumination, which varies over
at least 11 orders of magnitude, can be viewed as noise. To do this,
the statistics of the reflectance, surface orientation, and the
illumination will be use to calculate Bayes optimal filtering
methods. The characteristics of these ``filters'' will then be
compared to those inferred to be operating in low level vision.

The position is for 3 years and the wage is at research assistant A
scale.

Candidates should send a detailed CV together with the names and
addresses of two referee's to Dr Roland Baddeley, Laboratory of
Experimental Psychology, School of Biology, Sussex University, Falmer
Brighton BN1 9QG. More details can be found by emailing Roland
Baddeley on rjb@biols.susx.ac.uk
 
Closing date, January 28, 2000



This archive was generated by hypermail 2b29 : Fri Apr 07 2000 - 10:32:09 PDT