STATIONARITY INDEX FOR ABRUPT CHANGES DETECTION IN THE TIME-FREQUENCYPLANE

This paper presents a new method based on time-frequency representatio ns (TFR's) for detecting abrupt changes in non stationary noisy signal s. Stationarity indices (SI's), based on different distance measures b etween TFR's, are defined and a comparison of the performances is perf ormed using ROC curves and statistical properties. Two sets of signals are specially studied here: signals presenting very short segments an d nonstrictly stepwise stationary signals, For these signals, classica l methods based on autoregressive (AR) spectral estimators are often u nsuccessful.