It is difficult to extract signals under Low Signal Noise Ratio in biomedical signal processing. The elimination of movement artifact is really a bottleneck. A new solution for the movement artifact in pulse signal is proposed in this paper. According to pulse signal features, the signal is decomposed by using wavelet transform firstly. Then empirical mode decomposition (EMD) is applied to the wavelet coefficients in frequency band of useful signals, thus the signal and movement artifact can be distinguished effectively. Furthermore, the effectiveness of the proposed approach is verified by signal-to-noise ratio, energy ratio, cross correlation coefficient and power spectrum. This method can eliminate not only movement artifact, but also baseline wander and high-frequency noise. Thus, it provides an effective approach for the calculation of pulse rate and blood oxygen saturation.
Algorithms ; Artifacts ; Humans ; Pulse ; Signal Processing, Computer-Assisted