Minimal Cost-Path for Path-Based Distances

Distance functions defined by the minimal cost-path using weights and neighbourhood sequences (n.s.) are considered for the constrained distance transform (CDT). The CDT is then used to find one minimal cost-path between two points. The behaviour of some path-based distance functions is analyzed and a new error function is introduced. It is concluded that the weighted n.s.-distance with two weights (3 times 3 neighbourhood) and the weighted distance with three weights (5 times 5 neighbourhood) have similar properties in terms of minimal cost-path computation, while the former is more efficient to compute.