Three-dimensional skeletonization of elongated solids

Abstract A three-dimensional skeletonization algorithm using the tetradecahedral neighborhood in the face-centered cubic tessellation is described. The algorithm operates using table lookup logical transforms which lead to fast hardware implementations. this paper demonstrates the algorithm as applied to serial-section reconstructions of neurons from one of the nematodes. It is shown that use of the tetradecahedral neighborhood may lead to the production of “rings” which interfere with the skeletonizing algorithm. A procedure for detecting and deleting or modifying these “rings” is described.