Parallel and robust skeletonization built on self-organizing elements.

Kalmár, Zsolt; Marczell, Zsolt; Szepesvári, Csaba; Lörincz, Andras · Neural Netw · 1999

basic_science · Level V

Where this comes from

Abstract

A massively parallel neural architecture is suggested for the approximate computation of the skeleton of a planar shape. Numerical examples demonstrate the robustness of the method. The architecture is constructed from self-organizing elements that allow the extension of the concept of skeletonization to areas remote to image processing.