An improved adaptive signal segmentation method using fractal dimension

Analysis of non-stationary signal requires that it be segmented into piece-wise stationary epochs as many of the existing signals processing techniques are only applicable to piece-wise stationary signals. In this research, an adaptive segmentation approach is introduced that can automatically detect the positions of segments boundaries. In the proposed approach, after applying Savitzky-Golay filter on the original signal, the fractal dimension of the obtained signal is calculated in a sliding window. Then, segments boundaries are detected by considering fractal dimension variations. Performance of the proposed method is compared with an existing segmentation method using both synthetic signal real data. Simulation results indicate superiority of the proposed method in signal segmentation.