Sampling method of digital elevation model

The existing photogrammetry sampling methods involve DEM data collection, including contour line method, regular grid method, selective sampling method, progressive sampling method, profile method and mixed sampling method. If a function G( X) is sampled at an interval DX, the frequency part of the function higher than 1/( 2x) cannot be recovered by repeatedly sampling the data. There are three sampling methods with different viewpoints: sampling based on statistical viewpoint, sampling based on geometric viewpoint and sampling based on features.

Sampling from the statistical point of view: DEM surface can be regarded as a group of specific points (or sampling space), and there are two methods to sample this group (Yan Wang, 2006; Wei Yuchun, 2005).

Geometric sampling: DEM surface can be represented by geometric structure, which can be divided into regular form and irregular form according to its own properties, while the former can be subdivided into one-dimensional structure and two-dimensional structure. The irregular structure used for data point sampling is usually an irregular triangle or polygon. For regular structures, if they show regular characteristics in one-dimensional space, the corresponding sampling method is called profile method or contour method; A two-dimensional regular structure is usually a square or rectangle, or it may be a series of continuous equilateral triangles, hexagons or other geometric figures.

Feature-based sampling: DEM surface consists of a limited number of points, and the information contained in each point may change with the position of the point on DEM surface. From this point of view, all points on DEM surface can be divided into two groups: one group consists of characteristic points and lines, and the other group consists of random points. Feature points refer to surface points that contain more or more important information than ordinary surface points, such as peaks and valleys. Characteristic lines are lines formed by connecting characteristic points, such as ridge lines, valley lines, fault lines and structural lines. Feature points and lines not only contain their own coordinate information, but also implicitly express some information about their surrounding features. More importantly, if only the feature points and lines are collected for the whole surface, the main features of the surface can still be obtained. In this sense, sampling methods can be divided into selective sampling and non-selective sampling.