Pulse Signal Baseline Drift Removal Using Median Filter Method

In signal processing, a common challenge involves baseline drift in pulse signals and nonlinear one-dimensional signals. Baseline drift in pulse signals refers to temporal variations in the signal waveform, while in nonlinear one-dimensional signals it denotes spatial variations along the signal axis. To accurately analyze and extract signal features, it is essential to remove these baseline drifts. A widely adopted method is the median filter approach, which effectively eliminates baseline wander through a sliding window technique. The implementation typically involves: 1) Defining an appropriate window size based on signal characteristics, 2) Computing the median value within each window segment, 3) Subtracting the median values from the original signal to obtain the baseline-corrected output. This non-linear filtering technique preserves signal edges while removing low-frequency drift components, making signal features more distinguishable for subsequent analysis such as peak detection or feature extraction.