The medial axis transform (MAT) of a shape, better known as its skelet
on, is frequently used in shape analysis and related areas. In this pa
per a new approach for determining the skeleton of an object is presen
ted. The boundary is segmented at points of maximal positive curvature
and a distance map from each of the segments is calculated. The skele
ton is then located by applying simple rules to the zero sets of dista
nce map differences. A framework is proposed for numerical approximati
on of distance maps that is consistent with the continuous case and he
nce does nor suffer from digitization bias due to metrication errors o
f the implementation on the grid. Subpixel accuracy in distance map ca
lculation is obtained by using gray-level information along the bounda
ry of the shape in the numerical scheme. The accuracy of the resulting
efficient skeletonization algorithm is demonstrated by several exampl
es. (C) 1995 Academic Press, Inc.