Signal Characterization 1. · Definition for method selection
Stationarity (ADF, KPSS, running mean/var)
SNR estimation
Normality checking
Spectral characterization (PSD)
Result → determines the preprocessing method
↓
Non-stationary → EMD/Wavelet | Stationary → FIR/IIR | Low SNR → Signal Averaging | Non-stationary → Stationarity-based segmentation
Preprocessing Digital Post ADC
Detrending
Mean subtractionPolynomial fittingSpline detrendingHP filtre (dijital)
Signal Enhancement / Noise Reduction
Signal Averaging (ensemble · SNR ∝ √N)Wiener filtreSmoothing
Interference Suppression (Dijital)
Digital notch filtreAdaptive filtering (LMS / NLMS / RLS)Spectral subtraction
Artifact Removal
Filtering based
FIR (linear phase)IIR + filtfiltBaseline wanderPower line
Decomposition based
ICAPCAWavelet thresholdingEMD / EEMD / CEEMDANVMD
Template / Adaptive
Template subtraction (BCG)LMS / NLMS / RLS
Hedef artifact'lar
EOG · ECG · BCGEMG · MotionPower line · Baseline
Normalization / Resampling
Z-score · Min-MaxDownsampling · Interpolation
Segmentation
Event-locked (Epoch extraction)
Stimulus / trigger[-200ms, +800ms] pencereBaseline correctionERP · AEP · VEP · SSEP
Physiological event-based
R-peak → beat segmentationP / QRS / T izolasyonuBreath cycle (resp.)
Stationarity-based
Quasi-stationary partitioningFor non-stationary signalCharacterization'dan tetiklenir
Sliding / overlapping window → At Feature Extraction limit (below)
↓
Verification after preprocessing → Is SNR improved? Has the baseline been raised? Is the morphology disrupted?
Signal Characterization 2. · Preprocessing verification
SNR comparison (before/after)
Residual artifact control
Spektral analiz (filtre etkisi)
Signal morphology verification
If insufficient → go back to preprocessing step