Abstract :
Linear transforms of the characteristic functions of plane figures are averaged over the one parameter continuous group of dilatations (homothetic scale changes) to obtain new dilatation-invariant transforms. For a large class of kernel functions, the group averaged transforms act as corner detectors or edge detectors.
Keywords :
Artificial neurons, corner detection, edge detection, group averaging, linear transforms, pattern classification, plane figures, scale-invariant features, transformation invariants, threshold logic networks.; Detectors; Humans; Image edge detection; Kernel; Logic; Mathematics; Neurons; Pattern classification; Retina; Shape; Artificial neurons, corner detection, edge detection, group averaging, linear transforms, pattern classification, plane figures, scale-invariant features, transformation invariants, threshold logic networks.;
