SpecImage module for python

A general module for hyperspectral image analysis



Developed by Leonardo de Oliveira Martins at the University of Vigo and at the Imperial College London for the SAVVY Project.

Main Features

  • Import text files from WiTec and Matlab; saving to Pickle format
  • Plotting of spectral densities
  • Removal of noise and cosmic spikes from spectra
  • Interpolation and rescaling (normalization)
  • Image simplification/reduction
  • Background and baseline corrections
  • Endmember extraction and abundance maps
  • Modelling, clustering and classification

Noise and cosmic spikes

Noise removal based on autocorrelation function
Cosmic spikes are found by derivatives over pixels and wave numbers
showcase_pictures/a002_01_static.png

Plotting of distribution of spectra


OBS: the images represent distinct images

Interpolation

Savitzky–Golay filter
Moving average smoothing
Also
  • Linear interpolation
  • Spline interpolation
  • LOWESS regression

Prediction via interpolation
showcase_pictures/a001_05_spline_smooth10_static.png

Normalization


def rescale_01(self):
  # Linear rescaling to interval zero-one
  self.spc = np.array([rescale_spectrum_01(spec) for spec in self.spc],dtype=float)
  return self

def rescale_zero(self):
  # force series to start at zero, while maintaining range
  self.spc = np.array([spec - min(spec) for spec in self.spc],dtype=float)
  return self

def rescale_mean(self):
  # centralization to the mean
  self.spc = np.array([spec/np.mean(spec) for spec in self.spc],dtype=float)
  return self

def rescale_zscore(self):
  # Standardization of spectrum to a N(0,1)
  self.spc = np.array([(spec - np.mean(spec))/np.std(spec) for spec in self.spc],dtype=float)
  return self

Simplification of spectral matrix


Similar pixels replaced by 'exemplar' spectra

List of 'exemplars'

Background correction by PCA transform

  • Application of a PCA dimensionality reduction followed by inverse transform
  • Recreates original data but using only principal components
  • Can be applied over wave numbers or over spectra (pixels)
(1) Original data
(2) after PCA noise reduction
(3) after PCA pixel reduction

Baseline correction

Rubberband baseline
 Polynomial baseline
Piecewise rubberband baseline
 Asymmetric least squares baseline