An iterative nonlinear Gaussianization algorithm for image simulation and synthesis, (with J.-J. Lin and R. A. Levine), submitted for publication, 2001.

Abstract

We propose an Iterative Nonlinear Gaussianization Algorithm (INGA) which seeks a nonlinear map from a set of dependent random variables to independent Gaussian random variables. A direct motivation of INGA is to extend principal component analysis (PCA), which transforms a set of correlated random variables into uncorrelated (independent up to second order) random variables, and Independent Component Analysis (ICA), which linearly transforms random variables into variates that are "as independent as possible." A modified INGA is then proposed to nonlinearly transform ICA coefficients into statistically independent components. To quantify the performance of each algorithm: PCA, ICA, INGA, and modified INGA, we study the Edgeworth Kullback-Leibler Distance (EKLD) which serves to measure the "distance" between two distributions in multi-dimensions. Several examples are presented to demonstrate the superior performance of INGA (and its modified version) in situations where PCA and ICA poorly simulate the images of interest.

Get the full paper: gzipped PS file or PDF file.



Please email me if you have any comments or questions!
Go back to Naoki's Publication Page