A computer science perspective on the bendsimplification algorithm

Author: S. Herbert, M. Visvalingam, Mahes Visvalingam
Publisher: Cartography and Geographic Information Society (CaGIS)

ABOUT BOOK

The primary aim of this study was to evaluate whether the use of bends provides a better basis than point elimination for research on line structuring. These investigations were undertaken using Arc/Info 7.1.1. Comparative experimental results suggest that the algorithm may not be as widely applicable as the much simpler geometric filters, such as the Douglas-Peucker or Visvalingam algorithms. The paper therefore provides a brief review of these three algorithms. A more detailed conceptual and empirical evaluation of the bendsimplification system follows, highlighting some problems with implementing the system in Arc/Info. The paper then questions the value of over-coupling model- and image-oriented generalization processes within the black-box bendsimplification system. It suggests the type of parameters which could enhance the utility and usability of the Bendsimplify option within the Arc/Info (and perhaps also within the ArcView) environment and provides some pointers for further research. With respect to the main aim of the research, the evidence suggests that bendsimplification is less useful for line segmentation than Visvalingam's algorithm. Further research is needed to assess the value of the iterative bend elimination operator within bendsimplification

Powered by: