A cascade filter for pulse wave baseline drift elimination

Analyzing the pulse wave is an important method to diagnosis the status of human health. Generally the directly acquired signal contains varying degrees of baseline drift which is usually caused by breathing and physical movement, it mostly influences the accuracy of signal analysis and must be eliminated. This article presented a cascade filter framework to remove the baseline drift of pulse wave. The first stage zero phase filter (ZPF) roughly reduced the trend of the baseline drift. The second stage of the removed method considered the signal features. Through the pulse waveform reference points´ accurate location and cubic spline estimation, we completely eliminated the rest drift. Compared with the other methods, e.g. wavelet-based filtering, morphological filtering, empirical mode decomposition (EMD) filtering, the experimental result showed that the proposed method had not only better effect, but also small calculation cost, and the scheme was more suitable for actual application.