AUTOMATIC DENOISING USING LOCAL INDEPENDENT COMPONENT ANALYSIS
Peter J. Gruber, Fabian Joachim Theis, Ana Maria Tomé, Elmar Wolfgang Lang · University of Regensburg Publication Server (University of Regensburg) · 2004
We present a denoising algorithm for enhancing noisy signals based on local independent component analysis (ICA). We extend a noise reduction algorithm proposed by Vetter et al. [11] by using ICA to separate the signal from the noise. This is done by applying ICA to the signal in localized delayed coordinates. The components resembling noise are detected using estimators of kurtosis or the variance of the autocorrelation. This algorithm can also be applied to the problem of denoising multidimensional data like images or fMRI data sets. In comparison to denoising algorithms using wavelets or Wiener fllters the local PCA and ICA algorithms perform considerably better especially ICA algorithms which consider the estimation of higher order statistical moments like kurtosis.