关键词: 数字高程模型, 地形特征, 最小二乘法, 随机抽样一致性算法, 失真修复

Abstract: There is a certain degree of distortion in Digital Elevation Model (DEM) depicting terrain. It is necessary of recovering distorted DEM to improve the accuracy and fidelity. In this paper, a method of recovering distorted DEM of regular terrain under constraint of terrain feature is proposed by taking a flat terrain object as an example. Firstly, the DEM of flat terrain should be a flat structure according to terrain feature, and it can be described by a plane equation. Then the expression of plane can be estimated by parameter estimation algorithms, this paper selects least square method (LS) and random sample consensus algorithm (RANSAC) to estimate it. And the recovered DEM will be obtained based on the expression. The recovery results reveal that the method is effective and feasible, and the recovered DEM not only has a higher accuracy, but also can reflect the terrain feature fully and exactly. With massive grid points, both algorithms can achieve a great recovery effect. However, RANSAC algorithm has a better robustness than LS and RANSAC can better adapt to the mutations of elevation.

Key words: digital elevation model, terrain feature, least square method, random sample consensus algorithm, recovery