This matlab code provides a mechanism to construct a lookup table and
warp a digital image via the monopole map [1] or dipole map [2] of the
visual field.  There are two main functions: constructLogmap.m and
mapImage.m.  Detailed descriptions of the usage of these functions can
be obtained by typing

    help constructLogmap
        -or-
    help mapImage

at a Matlab command prompt.  To further illustrate the usage, three
demo scripts have been provided.  Type

    logmapSimple
      -or-
    perfectScaleDemo
      -or-
    mapComparisonDemo

at the Matlab command prompt to run the desired script.  These scripts
use images from the CV/CNS image database (a database of images available
for use under the Creative Common License).

The first script, logmapSimple, uses a small scale factor to produce the
output maps.  It will display the original image as well as the
monopole or dipole images. It can be run on any MATLAB supported image
and supports a variety of parameter choices.
(See 'help logmapSimple' for details.)

The second script, perfectScaleDemo, uses a modified form of the
algorithm described in [1] to obtain "perfect scale."  Perfect scale
means that a unit step taken from the center of the original image
maps to a unit step in the center of the warped image.  Thus, the peak
resolution of the warped image is identical to the resolution of the
original image.  When the warped image is mapped back to the
coordinates of the original image, the image resolution noticeably
falls off as a function of the distance from the image center.  This
is due to the lossy nature of the logarithmic compression.

The final script, mapComparisonDemo, demonstrates how the monopole
mapping relates to the dipole mapping.  The monopole approximates the
topographic structure of the central 20 degrees of the primate visual
field, whereas the dipole describes the full field (somewhere in the
vicinity of 60-100 degrees).

Note: These scripts may take several seconds to run.

INSTALLATION

Add this directory to your matlab path.  Alternatively, just <cd> to 
this directory in matlab and run the code.

  References:
  ===========
 
 [1] Alan S. Rojer and Eric L. Schwartz. Design considerations for a 
     space-variant visual sensor with complex-logarithmic geometry. 
     In Proceedings. 10th International Conference on Pattern 
     Recognition, volume 2 of International Conference on Pattern 
     Recognition, pages 278-285, Atlantic City, NJ, June 16-21 1990. 
     Int. Assoc.  Pattern Recognition, IEEE Comput. Soc. Press.

 [2] Mukund Balasubramanian, Jonathan Polimeni, and Eric L. Schwartz. 
     The V1-V2-V3 complex: quasiconformal dipole maps in primate 
     striate and extra-striate cortex. Neural Networks, 
     15(10):1157-1163, 2002.