Extraction of characteristic points and its fractal reconstruction for terrain profile data

The main objective of this paper is to study reduction rate of 2D DEM (digital elevation model) data profile after data reduction by the Douglas-Peucker (DP) linear simplification method and by fractal interpolation to show original terrain reconstruction. In this paper, two-dimensional data of measured geographic profiles are taken as the study object, by using the DP method and the improved Douglas-Peucker (IDP) method to reduce data. Its aim is to retain spatial linear characteristics and variations, then take reduced data points as basic points and use the random fractal interpolation approach to add more data points up to the same as the original data points, in order to reconstruct the terrain, and compare the experimental data with the random point extraction method addressed in related literature. This paper uses tolerance calibration to generate different reduction rates and utilizes four types of evaluation factors, statistical measurement, image measurement, spectral analysis and elevation cumulative probability distribution graph, to make a quantitative analysis of profile variation. The study result indicates that real profile elevation data, manipulated with varied reduction approaches, then reconstructed by means of fractal interpolation can produce data points with a higher resolution than those originally observed, thereby the reconstructed profile gets more natural and real details.