One-dimensional smoothing using extreme envelope average

In this paper, a new de-noising technique called extreme envelope average (EEA) is presented. This technique is related to the energy reduction of the noisy data throughout the successive averages between the maximum and minimum extreme envelopes. The basic principle is simple and is based on the central tendency of the extreme averages after a specific number of sets of the process. Important structures of the signal, representing low-frequency components, are maintained. It is a very fast convergence method, requires a unique noisy data, and presents satisfactory results. There are no constraints about the linearity, stationary or harmonic content in relation to of the test signal, that make it a useful approach for signal that presenting abrupt changes along of its curvature. And, it can also be used for non-equally-spaced data. The present study is limited to signals numerically corrupted by white Gaussian noise with zero mean.

► The study was done to signals which have different curvature characteristics and affected by Gaussian random noise.
► The accuracy of the results are directly related to number of samples, noise levels and type of signals.
► The time of convergence is influenced by the number of samples.
► The adopted value to the stopping criterion based on MAC affects the number of convergence iterations.