BSc thesis handed in on impact of spectral and spatial resolution on classifications

BSc thesis handed in on impact of spectral and spatial resolution on classifications

written by Martin Wegmann

October 26, 2017

Marius Philipp handed in his B.Sc. thesis on “Spectral and spatial resolution of remote sensing data and its impact on classification processes – A case study of the Bavarian Steigerwald”. From the abstract: Remote sensing data enables a highly potent and cost efficient way for extracting thematic information such as land cover types and their distribution. A common area of application is therefore the creation of land cover and land use maps. But since this is a complex decision-making process, a successful execution of this task can prove to be rather difficult. Choosing suitable imagery in terms of its spatial and spectral resolution can highly affect the accuracy results of the finished product and should therefore be carefully considered beforehand. In order to assist ecologists by giving them a reference point for choosing the best suited data, this study contains comparisons of several remotely sensed data sets with different spatial and spectral features. Four individual classification algorithms are applied in a multi-level classification approach on numerous scales, which is designed to be increasingly complex with each classification scheme. Results of this study reveal, that higher spatial and spectral resolutions do not always generate more accurate classification results. It is shown, that satellite imagery with a medium spatial resolution tend to create better land cover maps compared to high resolution images. Furthermore,  the individual classification algorithms tend to strongly affect the quality of the classification products as well, which creates another import decision point that will be discussed in this study.

 

supervisor: Dr. Martin Wegmann in collaboration with the Steigerwald research station.

you may also like:

Our urban Earth Observation research is part of CAIDAS

Our urban Earth Observation research is part of CAIDAS

Hannes Taubenboeck is also linked to the CAIDAS, which is the Center for Artificial Intelligence and Data Science of the Julius-Maximilians-Universität (JMU) Würzburg which aims to develop AI methods, new research topics and the transfer into applications. This is...

Open short term position in Remote Sensing

Open short term position in Remote Sensing

Open short-term position (100%) in Remote Sensing Monitoring status and changes of the earth is essential for a sustainable management of our limited natural resources but also to ensure the functionality of our infrastructure. For a project dealing with the...