A New Adaptive Signal Segmentation Approach Based on Hiaguchi's Fractal Dimension

In many non-stationary signal processing applications such as electroencephalogram (EEG), it is better to divide the signal into smaller segments during which the signals are pseudo-stationary. Therefore, they can be considered stationary and analyzed separately. In this paper a new segmentation method based on discrete wavelet transform (DWT) and Hiaguchi’s fractal dimension (FD) is proposed. Although the Hiaguchi’s algorithm is the most accurate algorithms to obtain an FD for EEG signals, the algorithm is very sensitive to the inherent existing noise. To overcome the problem, we use the DWT to reduce the artifacts such as electrooculogram (EOG) and electromyogram (EMG) which often occur in higher frequency bands. In order to evaluate the performance of the proposed method, it is applied to a synthetic and real EEG signals. The simulation results show the Hiaguchi’s FD with DWT can accurately detect the signal segments.