                                                   Jeff Kuo
                                                   11/16/88



          Vision Seminar Readings Summary/Critique

B.K.P. Horn, Understanding Image Intensities

The process of image formation is explained in an attempt to understand
images.  Reflectance functions are defined and representations in
gradient space introduced.  The differences between specular and matte
surfaces in gradient space are described.  Edge profiles allows
interpretations of convexity or obscuration.



Alex Pentland, Shape Information From Shading:
	A Theory About Human Perception

The reflectance function will be approximately linear if the average
illuminant is more than plus or minus 30 degrees of the viewer's
position.  If this is so, it can be shown that the image intensity
surface is a linear function of the height of the surface.  Depth can be
easily recovered in this case.  It is shown that this simplifying
assumption is likely to be what happens in biological system, i.e. the
depth that we perceive by shading cues is based on the assumption of a
linear reflectance function.  Whenever the algorithm makes a mistake
because the quadratic terms are not negligible, the subjects make the
same mistake in their perception of depth.  A biological mechanism based
on the Gabor filter qualities of the cells between the retina and the
striate cortex is proposed that would allow the extraction of depth
information from shading using on the same sort of operations performed
by the algorithm.  (Taking Fourier transforms, etc.)

