A three-dimensional Douglas-Peucker algorithm and its application to automated generalization of DEMs

This article reviews the formation of the 3-D Douglas-Peucker algorithm on the basis of analyzing the nature of the 2-D algorithm and further studies the application of the 3-D method to the automated global generalization of digital elevation model (DEM). Compared to the general 3-D Douglas-Peucker algorithm put forward by the authors in the previous publication, further improvements and expansion of the algorithm have been included in this article, namely, (1) for randomly distributed points, 'loneliness index' of the current point has been taken as the dynamic weight factor of the point-plane distance, so as to improve the selection of the feature points; (2) aiming at the mass volume of the regular square grids, which forms the majority of the DEM data; three measures have been suggested in this article for the improvement of the efficiency of the automated generalization. Experiments have proven that these measures can greatly heighten the efficiency of the DEM generalization with satisfactory results and have offered us the practical possibility of on-the-fly global generalization of DEM with a huge volume of data.