gh@utai.UUCP (Graeme Hirst) (10/04/84)
. ARTIFICIAL INTELLIGENCE SEMINAR DEPARTMENT OF COMPUTER SCIENCE UNIVERSITY OF TORONTO Skewed symmetry: Its physiology and technique Chris Brown University of Rochester Tuesday 9 October, 3pm, Sandford Fleming 1105 ABSTRACT Symmetric planar shapes viewed under orthography from an arbitrary position have a (skewed) symmetry axis and a skew (transverse) axis. If thye are known, th plane's orientation may be constrained to a one-dimensional locus in the spherical space of orientations. The second moments of skew-symmetric image shapes constrain the axes to a one-dimensional locus in the toroidal space of axis-angle pairs. Then finding skew and symmetry axes involves searching this locus for a pair of angles that in fact specify satisfactory axes. In a quantized image, this calls for robust symmetry evaluators. The analytic constraints will be developed, several symmetry evaluators considered, and experimental results shown. -- \\\\ Graeme Hirst University of Toronto Computer Science Department //// utcsrgv!utai!gh / gh.toronto@csnet-relay / 416-978-8747