Signal Smoothing Based on Centralization of Local Extremes

In this paper, a new iterative 1-D smoothing technique is proposed. The core idea behind it is based on the centralization or compression of the signal-noise amplitudes and outliers by the repositioning of the local extremes, maxima and minima, inside the central region of the noisy data where the required noiseless information is known to be. The technique presents a very straightforward implementation without requiring any special mathematical functions, at the same time keeping important and complex signal components, such as jumps and edges. Simulated and real data are provided to illustrate the effectiveness of the method.