scPrepareFilters

 Prepare the spatial filters for S-CIELAB opponent blurring

   [filters,support,params]  = scPrepareFilters(params)

 Create three pattern-color separable filters according to the Poirson &
 Wandell 1993 fitted spatial response. The filters are each weighted
 sum of 2 or 3 gaussians.  They are in the cell array, filters.

 The params structure defines several properties of the features.
 The two important slots are

 sampPerDeg -- filter resolution in samples per degree of visual angle.
 Minimum sample rate is recommended as 224 samples per degree of visual
 angle.  Code issues a warning of the sample rate is smaller than 100.

 dimension:  Specifies whether the created filters should be 1-D or 2-D.
 These are obscure conditions (normally 2 is used) and we need to improve
 the comments here.

   dimension = 1: generate the linespread of the filters;
    This is useful for 1-d image calculations, say for theoretical 
    work.

   dimension = 2: generate the pointspread of the filters;
    This is useful if you just want to create an image of the filters

   dimension = 3: generate the pointspread in the form that can be used by 
     separableConv.  The result is a set of 1-d filters that can be applied
    to the rows and cols of the image (separably).  This is not used in
    ISET, but it may be used elsewhere.

 The filters are a cell array.  The support defines the spatial support in
 terms of degrees of visual angle.

 Example:
   params.deltaEversion = '2000';
   params.sampPerDeg    = 145;
   params.imageFormat   = 'LMS';
   params.filterSize    = 145;
   params.dimension     = 2;
   [filters,support]    = scPrepareFilters(params);

   figure;   % Units are degrees of visual angle
   subplot(1,3,1), mesh(support,support,filters{1}); colormap(hsv(256)); 
   subplot(1,3,2), mesh(support,support,filters{2}); colormap(hsv(256)); 
   subplot(1,3,3), mesh(support,support,filters{3}); colormap(hsv(256)); 
 
 In 1996, Xuemei Zhang used this routine
   [k1, k2, k3] = separableFilters(params.sampPerDeg,2);
   figure(1); clf
   subplot(1,2,1), mesh(support,support,filters{3})
   subplot(1,2,2), mesh(support,support,k1)

 The values are the same
   figure(1); clf, mesh(support,support,k1-filters{1})
   figure(1); clf, mesh(support,support,k2-filters{2})
   figure(1); clf, mesh(support,support,k3-filters{3})

 Copyright ImagEval Consultants, LLC, 2003.