Determining grain-size distributions from aerial photographs using eCognition

The definition of the surface particle-size distribution in gravel- and cobble-bed streams is relevant for a number of analyses regarding sediment transport, hydraulics, and streambed monitoring. Yet, particle-size distribution measurements by fieldwork are cumbersome and time-consuming.

Earlier studies have shown that it is possible to approximate particle-size distribution from the analysis of photographs using a special tool of Matlab.

The MSc-thesis will test if sufficiently accurate particle-size distributions can be automatically derived using the software eCognition:

  1. from aerial photographs
  2. from terrestrial photography

With sufficiently accurate we mean: with similar accuracy as manual measurements.

The outcome of the Master thesis shall be a verification/falsification of the hypothesis that "Particle-size distribution can be automatically derived from photographs using the software eCognition". In the case of verification of the hypothesis, an algorithm (a number of algorithms) is expected as outcome of the thesis.

Data basis:
Existing at NGI:

  • Photographs (airborne and terrestrial) of gravel- and cobble accumulations in different river beds in Norway
  • Ground-truth data (i.e. manually measured particle-size distribution) from a limited number of field sites.
  • eCognition software license

Supervision can be in Norwegian, English or German. Master thesis will have to be written in English.

Published Nov. 3, 2017 10:38 AM - Last modified Nov. 3, 2017 11:10 AM

Scope (credits)