University of Hull logo

Bloch's function D for weighted effective areas : impact of tuning parameters

Visvalingam, Maheswari


Abstract When Visvalingam’s algorithm was first presented, it was noted that it could be driven with any metric to suit different purposes, types of features and degrees of generalisation. It was illustrated with the concept of the effective area (EA). The intelligent application of metrics with and without weighting requires the prior segmentation of polylines with heterogeneous shapes into geometrically similar sections. Automatic line segmentation remains a research challenge. Meanwhile, different weighting functions have been used with different types of lines as noted in this paper. EA has a tendency to output spiky lines. Attempts have been made to give less weight to acute angles and favour obtuse angles. This yields visually more pleasing lines but has a tendency to chop elongated features, such as streams. Bloch’s function D (which implements the Visvalingam/ weighted area option in recent versions of Mapshaper) was chosen to investigate whether the parameter values could be tuned to improve the output. The tweaking of parameter values can give unexpected and unpredictable results. The results from a systematic approach to tweaking the parameters are presented here. These suggest that the chopping effect may be exacerbated by the use of inappropriate parameter values. By fine tuning Bloch’s function D, it was possible to derive pleasing smaller scale representations of convoluted coastlines with a complex network of creeks using under 5 percent of the original points. Although the output is not identical to manual generalisations of the same coastline, they may be acceptable (at this stage of research) for this type of coastline. Function D is not useful for generalising unsegmented coastlines, such as the 1: 50000 coastline of Humberside.

Book Type Authored Book
Journal Explorations in digital cartography
Peer Reviewed Not Peer Reviewed
Book Title Bloch's function D for weighted effective areas : impact of tuning parameters
Keywords Mapshaper


Downloadable Citations