A parallel MSF algorithm for planar graphs on a mesh and applications to image processing

Abstract

The author presents an efficient O(n) parallel algorithm for finding a minimum-cost spanning forest (MSF) of a weighted undirected planar graph with n/sup 2/ edges, on an n*n mesh-connected computer. He also obtains efficient MSF-based O(n) algorithms for several application problems in image processing. In particular, he shows that an MSF can be used to obtain more efficient and elegant O(n) algorithms for the 'k-width connectivity' problem and the 'optical clustering' problem.<>