Imagine we are surveying the topography of the surface of the earth through a remote sensing instrument located on a satellite in space. This instrument is highly accurate in many situations, but has the special characteristic that it cannot accurately calculate the height (above sea level) for locations on earth, but it can calculate the relative height differential between two neighbouring surface points very accurately, as long as the height differentials are not too large. When the differentials are too large we will observe a measurement error very similar to a phenomenon known as "phase wrapping", where we will get a totally wrong differential measurement. The threshold for when a measurement turns unstable depends on the frequency of the signal beamed from the satellite, where the choice of signal frequency is a compromise between accuracy, sensitivity to noise, distance to target, etc.
The idea is to use Machine Learning in order to decide between stable / unstable signals.
Further work on the project described at Uppsala (in collaboration with Victor Aarre (Schlumberger):
[Necessary] TDT4195 (Visual Computing Fundamentals)
[Preferably] TDT4230 (Graphics & Visualization), TDT4265 (Computer Vision), Machine Learning.
Dataset to be used: Will be provided by Schlumberger
Advisors: Liyuan Xing (NTNU), Victor Aarre (Schlumberger), Theo Theoharis (NTNU).
If interested, please contact Liyuan Xing, email@example.com