Relief Automated Generalization Based On The Improved Three Dimensional Douglas-Peucker Algorithm

Relief Automated Generalization is the foundation for people to observe, express and analysis the geographic phenomenon multi-dimensional, and it is the core work in constructing the multi-scale geographic database. Massive researches have been done in the theory and the method of the relief automated generalization. Three dimensional Douglas-Peucker algorithm can be use to get the different detailed relief feature points based on the three dimensional discrete points and then can achieve the relief automated generalization. This paper firstly improved the Three dimensional Douglas-Peucker algorithm in the basic plane and the algorithm direction, then study the correlation between the generalization level and the threshold value based on the DEM of the typically area in Loess Plateau, according to this, formed the method of get the best threshold value from the initially threshold value experiment in relief generalization. Based on the method and the improved Douglas-Peucker algorithm, achieved the relief generalization from the 1:10000 DEM to the 1:50000 DEM in the typically area in Loess Plateau. Finally, evaluate the generalize result of the 1:50000 DEM in elevation accuracy and terrain parameter according to the original 1:50000 contours and 1:50000 DEM. The result shows that the improved Douglas-Peucker algorithm can extract the overall relief feature points effectively, the generalize result of the 1:50000 DEM has a high accuracy, and achieve the fist level or second level of the accuracy criteria of national 1:50000 DEM, it's terrain parameter frequency spectrum also has a highly correlation with the terrain parameter frequency spectrum derived from the national 1:50000 DEM. Relief automated generalization based on the improved Douglas-Peucker algorithm avoid the shortcoming of recognize the DEM as the digital image, which depart the DEM from its terrain character. It abides by the rule of keeping the important features and neglecting the small features in relief generalization form extracting the relief feature points, and it comes to a good generalization result. To some extent it is the improve and perfect of the relief automated generalization based on DEM, and A case study in the Loess Plateau, it established the foundation for the Loess Plateau relief generalization and for the research on the digital Loess Plateau.