Unsupervised Detection of Topographic Highs

Introduction

The automatic detection of topographic highs from DEM can be applied in a wide range of the geoscientific disciplines such as remote sensing, topography, geology, geoarchaeology, oceanography and others. In this work, an unsupervised isocontour based segmentation method is proposed, that is applied on the detection of topographic highs with arbitrary basal shapes on Digital Elevation Models (DEMs).

Methodology

Figure 1.Scheme of the main steps of the VOLEI method

The proposed method of topographic high detection is based on the volume expansion of isocontours, which provides a series of isocontour based segmentation maps for decreasing altitude levels. It holds that the more challenging problem is the detection and discrimination of the topographic highs with complex shape that are close together.

Figure 2. (a) The automatically selected tops (white plus) (b) The maximum expanded region Rν(rmin) for the six selected tops. The maximum altitude (in m) and the rmin (m) are also provided for each detected topographic high, shown in different color. (c) The final results of the VOLEI method projected on Giouchtas DEM.

• The ambiguity of the highs' boundaries is efficiently solved by suitable isocontours that are derived as solutions of a probability based optimization problem, based on the volume evolution of an isocontour starting from the top and gradually growing, as decreasing the altitude level of the isocontour.
• The order of the topographic high detection is given by the inclusion tree that represents the enclosure relationships among the isocontours.

Downloads

• You can download the matlab code of VOLEI method proposed in [1].
• The Synthetic Dataset of 180 DEMs (300×300 grid cells) using in [1].

Related Publications

[1] C. Panagiotakis and E. Kokinou, Unsupervised Detection of Topographic Highs with Arbitrary Basal Shapes Based on Volume Evolution of Isocontours, Computers & Geosciences, vol. 102, pp. 22-33, 2017.