A fast 3D seed fill algorithm avoiding redundant seed searching by 2 ½ D linked lists

Abstract

This paper presents a 3D seed and flood algorithm that uses a 2 ½ D linked list to avoid the redundant seed searches that appear in the current 3D seed and fill algorithms. The linked lists records a set of shadows cast by the filled spans. We maintain the linked lists and com-pare the extent of the current span with the extents in neighboring spans to minimize refer-ences to the voxels. The count for filled voxels is corrected and verified by the compari-sion with the Oikarinen’s 3D algorithm. Verification of our algorithm and the Feng’s algo-rithm by artificial and measured volumes demonstrates the effectiveness of our method in eliminating redundant seed searching and thus improving the efficiency.