New domain block partitioning based on complexity measure of ECG

This paper introduces a new domain block partitioning scheme for a nonlinear iterated function systems (NIFS) compression of electrocardiogram (ECG) signals, based on their complexity measure. The idea behind the scheme is based on the multifractal characteristics of the ECG signal. The partitioning is intended to reduce the time-consuming inverse problem in fractal compression. The proposed technique gives computational complexity of O(N) for a time series with length N. The segmented NIFS achieves a compression ratio of 8.8:1 under a distortion error of 5.8%, as compared to that of 6.0:1 obtained by Olen and Narstad´s orthogonal transform.