Coastline matching via a graph-based approach

Abstract This paper studies the problem of unsupervised detection geometrically similar fragments (segments) in curves, context boundary matching. The goal is to determine all pairs sub-curves that are similar, under local scale invariance. In particular, we aim locate existence a section (independent length and/or orientation) second curve, first as indicated by user. proposed approach based on suitable distance matrix two given curves. Additionally, objective function capture trade-off between similarity common sub-sequences and their lengths. algorithm minimize this via an efficient graph-based capitalizes Dynamic Time Warping compare subcurves. We apply technique geometric matching coastline pairs. application crucial for investigating forcing factors related evolution. method was successfully applied global data, yielding bipartite graph with analytical point-to-point correspondences.