Spectrum-Based Fractal Analysis of Myoelectric Signals Using Piecewise Statistically Self-Affine Power-Laws

Abstract

n this paper we present a novel set of statistically self- affine power laws and an algorithm for parameter estimation of a piecewise power law combination. The piecewise combination is applicable to irregular power spectral densities which do not follow the classic form of strict statistical self-affinity. The piecewise modeling also enables local analysis with variable magnification factors, which is very informative about the spectral distribution of the texture. Results of an experiment on simulated myoelectric signals are also presented. In this experiment, two conditions in which a single power law results in large errors are investigated. The results show that extension of the modeling to a piecewise combinational approach improves the accuracy and results in a better representation of the power spectrum. The results also show a great potential for applications of this approach to a wide variety of bio-signals with a multi-fractal behavior, which is very close to combinational mono-fractals in texture.