Last update: Nov. 28, 1998
Publications by Ugo Montanari et al. on shape symmetry elicitation :
BibTeX references.
G. Levi and U. Montanari
Information and Control, v.17(1), pp.62-91, August 1970.
U. Montanari
Jour. of the ACM, v.16(4), Oct. 1969, pp.534-549
The "propagation of the figure contour is simulated analytically".
An intrinsic coordinate system" and its "singularities" are defined.
A "simulation of the propagation process is performed on a discrete plane"
"The wavefront is composed of straight-line segments and arcs of circles, while the skeleton consists of straight-line segments and arcs of parabola".
"A successive wavefront can be obtained, according to Huygens' principle".
"A wavefront is a sequence of extended elements" :
Introduces "breakpoints" which are intersections of 2 or more non-adjacent extended elements.
A "final breakpoint" is "created by the coincidence of 3 (or more) "suture points of a vanishing wavefront" (that is like a 4th-order shock it seems).
An "intermediate breakpoint" is "created by the coincidence of 2 suture points of a nonvanishing wavefront" (this is a generic node: two branches coming together create a new one).
An "initial breakpoint" is "not created by the coincidence of suture points and "at least one of the 2 intersecting extended elements is an arc of circle" (this is the 2nd-order shock, a neck).
Some "special cases" are considered, which include the 3rd order shock (only 2 parallel sls are mentioned ... missing the 2 ac case?).
Follows a data-structure for the wavefront, which includes the parametric equation of every indefinite extended element, etc.
He then considers properties of breakpoints; in particular, the characteristic angles obey simple equations.
2 important theorems are given.
Theorem 1 : a breakpoint is either final, intermediat or initial
Theorem 2 : the total number of non-final breakpoints (i.e. initial - 2nd order shocks - and intermediate - tree nodes -) can be directly deduced from the initial wavefront (!!) (a formula is given)
An algorithm is detailed.
A formula for the outgoing speed of the "leaving branch" at an intermediate breakpoint is given (function of the speeds of the 2 incoming branches).
Then, 3 types of skeleton branches are considered:
Montanari does not consider 2 ac making-up a curved cylinder ... did he miss this (particular case) ? I guess he only considers generic cases ... although this is not clearly explained) the skeleton branch is an sls; speed function is given.
Ugo Montanari
Jour. of the ACM, v.15(4), Oct. 1968, pp.600-624
Page created & maintained by Frederic Leymarie,
1998.
Comments, suggestions, etc., mail to: leymarie@lems.brown.edu