00100	Neurophysiological studies suggest that special cells in the visual
00105	cortex of mammals extract edges.  A priori, intensity edges
00110	occur at discontinuities in surface slope or
00115	reflectance.  Images are patchworks of areas of obscuring objects.  Some
00120	edges correspond to boundaries between objects, some to interior edges
00125	of objects, others to surface markings.  All signify features of potential
00130	interest, but there is no direct connection between intensity boundaries and
00135	spatial interpretations (which are our perceptual target).  Thus, it is
00140	possible to find intensity edges on a local basis.  It is not possible
00145	to find the "meaningful edges" (eg for a line drawing) on a local
00150	basis, since those are chosen for spatial semantics and require a global
00155	interpretation of the scene.  Nor is it easy to find intensity edges;
00160	present techniques require extensive calculation and are not really
00165	adequate.  We expect that performance approaching human visual acuity
00170	inherently requires great computation cost, and that substantial
00175	economies in computation are possible only with degraded performance.
00180	For most problems of interest, however, those
00190	reduced abilities are presumably adequate.  
     

00100	Edges are discontinuities along a smooth curve, ie locally straight.  
00200	Edges can be distinguished from spots only in a two dimensional context.
00300	Thus local edge operators must be two-dimensional (over a small area)
00400	and directional.  Mammalian edge operators, Hubel-Wiesel cells, are directional;
00500	there are cells for many directions (angular resolution ??) and
00600	corresponding to retinal positions (spatial resolution and guess as
00700	to how many).  Presumably they are more dense
00800	near the fovea.  However it is not clear that these cells are
00900	distributed proportional to retinal cells.  Teleologically, straight
01000	line approximations to edges are less useful over large visual angles.
01100	Physiological evidence shows many more retinal cells than ganglion
01200	cells connecting them to the brain.  Only a portion of the fovea appears
01300	to be connected one-one to the visual cortex. It is known that shape
01400	discrimination is crude in the periphery of the field of view.
02000	What is done with the outputs of cells which are not directly connected
02100	to visual levels of the brain?  Do ganglion cells multiplex spatial
02200	information in the time domain, or is there primitive feature extraction
02300	in the retina?
02400	
02500	Hueckel [] created a local edge operator for computers, which computes
02600	directional moments and solves for a best edge.  The signal is primarily
02700	determined by the gradient, and thus the operator has an unfortunate
02800	sensitivity to smooth shading.
02900	
03000	Local edge detection as described here is a low-level operation in that
03100	no information from higher levels is used.  The output of local edge
03200	operators does provide a context for other operations.  
03300