A constant-time algorithm for computing the Euclidean Distance Transform on reconfigurable meshes

Abstract

The Euclidean Distance Transform (EDT) is a basic operation in computer vision, pattern recognition, and robotics. Fast computation of the EDT is needed since most of the applications using the EDT requires real-time computation. The authors propose a new and fast parallel algorithms for computing the EDT problem on the reconfigurable VLSI mesh model. It is shown that the VLSI complexity of the algorithm described is the same as those of other sequential algorithms and the mesh algorithm, while it uses much less time. To authors' best knowledge, this is the first constant-time EDT algorithm on any parallel computational model.