Bayesian and Variational Methods for Discontinuity Detection: Theory Overview and Performance Comparison

摘要: Discontinuity detection is of great relevance at different stages the processing and analysis geodetic time-series data. This paper essentially a review two possible methods. The first method follows stochastic approach exploits Bayesian theory to compute posterior distributions discontinuity parameters. epoch amplitude are then selected as maximum posteriori (MAP). second variational based on Mumford Shah functional segment detect discontinuities. Whereas original formulation was developed in continuous form, discrete approaches also available presenting some interesting connections with robust regressions. Both methods have been applied identify occurrence cycle-slips GNSS phase measurements. Simulated real data processed compare performance evaluate pros cons approaches. Results clearly show that both can successfully cycle-slips.