Curve registration (time warping)

Aligns waveforms by estimating and removing phase variation. Uses landmark registration or continuous registration.

Usage

registerCurves(
  x,
  method = c("continuous", "landmark"),
  landmarks = NULL,
  template = NULL
)

Arguments

x

A PhysioExperiment object or matrix (time x observations).

method

Registration method: "landmark" or "continuous".

landmarks

For landmark method, matrix of landmark times (landmarks x obs).

template

Template curve to align to. If NULL, uses mean.

Value

A list containing:

registered

Registered waveforms

warping

Warping functions

template

Template used for registration

Details

Phase variation (timing differences) can obscure amplitude differences in biomechanical data. Registration separates phase and amplitude variation.

References

Ramsay JO, Silverman BW (2005). "Functional Data Analysis." 2nd ed. Springer.

See also

fPCA() for functional PCA after registration, plotGaitCycle() (in PhysioMoCap) for plotting registered gait waveforms.

Examples

# Simulate data with phase variation
set.seed(123)
t <- seq(0, 100, length.out = 100)

data <- sapply(1:20, function(i) {
  phase_shift <- rnorm(1, 0, 10)
  t_shifted <- t + phase_shift
  sin(2 * pi * t_shifted / 100) * 30 + rnorm(100, 0, 2)
})

pe <- PhysioExperiment(assays = list(values = data), samplingRate = 100)
reg_result <- registerCurves(pe, method = "continuous")