Real life signals are mostly non-stationary, and the most interesting information they carry is in their non- stationary characteristics (the beginning or the end of an event, drifts, transient faults). In this work, we are extending Saito's algorithm for noise reduction and signal compression to non-stationary signals. This extension is achieved by adaptively segmenting the non-stationary signal in such a way that each segment of the signal behaves like a stationary signal. Our results show that this adaptive segmentation improves the noise reduction and the compression of the signal.
展开▼