英语论文网

1. Introduction
Digital elevation models (DEMs) have fundamentally changed the way we perceive elevation information. In the late 1990s, the emer-gence of high resolution (~ 1 m) elevation data allowed the explora-tion of our environment and its morphology with an unprecedented level of detail, making new applications possible (e.g. the study of micro-faults on cliffs). The visual analysis of a shaded DEM efficiently supports the detection of a great amount of features at various scales. Nowadays, earth-Science experts such as geologists and geomorphol-ogists use high resolution DEMs to visually assess geomorphological features. Thanks to the finer description offered by this new and rich source of information, the study of visual perception, i.e. the vis-ible phenomena or relevant structures, has evolved considerably. In-deed, it is now possible for instance to visually analyze a hillside and its details (elements of only a few meters) Since geological phenomena are composed of different nested topographical features, a multiscale approach is essential in geomorphological analysis. Klinkenberg (1992) suggested that a phenomenon fits over scales and that its features are nested in discrete scale intervals, leading to a strong correlation between features and phenomena (Mark and Aronson, 1984). In human vision, the neural network is able to distinguish specific features in relation to a corresponding scale ( Marr, 1982), as well as to carry out a multiresolution analysis.

1。引言

数字高程模型（DEM）已经从根本上改变了我们对高程信息的方式。在20世纪90年代后期，高分辨率显（~ 1米）的高程数据允许我们的环境，以前所未有的详细程度及其形态的考察，使得新的应用成为可能（例如微断层崖的研究）。一个阴影DEM有效地支持在不同尺度的大量特征检测的可视化分析。如今，地球科学专家如地质学家和地貌找到使用高分辨率的数字高程模型直观地评估地貌特征。多亏了更精细的描述所提供的新的和丰富的信息来源，视觉感知的研究，即可见吸收现象或相关结构，具有很大的。事实上，现在有可能为实例直观地分析一个山坡和它的细节（只有几米的元素）由于地质现象是由不同的嵌套的地形特征，多尺度方法在地貌分析是必不可少的。克氏（1992）表明一个现象，适合尺度，其特点是嵌套在离散尺度区间，导致特征和现象之间的强相关性（1984标记和阿伦森，）。人类的视觉，神经网络能够区分与相应规模的特定功能（马尔，1982），以及进行多分辨率分析。

As suggested byMarr (1982), our visual system is probably linked to tu ne d c ells or, i n o t her wo rd s, it h as s pe ci fic frequency intervals to which it is sensitive. Therefore, computer systems and the visual representations we make of processed data should reflect this multiscale nature. Nevertheless, in most current systems, information is perceived like a static image. There are two ways to interpret inform ation in an image: either we know what the image contains and we focus on retrieving this information using the most appropriate technique; or we do not know what the image contains and we use a more general method to identify and differentiate relevant content in the data. The latter applies to DEM analysis. Often, a given DEM contains speci fic information like a geological phenomenon, but the identi fication of its components is not straightforward. Consequently, topographical features have to be classified according to their representative scale. In ot her w or d s, i t i s n ec es sa ry to find the best correlation between a cer-tain level of generalization (of the DEM) and the scale of a particular fea-ture of the topography. This process is called a multiscale analysis of structural topographical features (Marceau and Hay, 1999).

High resolu本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。