perm filename CURVE[DOC,TOB] blob
sn#154249 filedate 1975-04-04 generic text, type C, neo UTF8
COMMENT ā VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002 design criteria for curve organization
C00004 ENDMK
Cā;
design criteria for curve organization
1. should operate in presence of large noise and heavily textured
regions (should be able to find outlines in non-textured areas,
without blowing up). The paradigm: random points are 50 times points
lying on curves. Pairs with a noise point are 2500 times as frequent
as pairs of curve points.
2. should span gaps locally
3. sensible combinatorially in view of 1 and 2
4. should be optimal with respect to contrast sensitivity.
1. linearize around the current fit in a worm-type curve fitting
2. how much better do we do by squeezing as far as possible
on global guesses for curves
3. consequence: relation between density of curve points and length
for constant detectability