Color Component Prediction Based on Damped Modified Independent Component Analysis

Abstract

Aiming at the problems of negative components and insufficient eigenvectors in the spectral space built by means of principal component analysis (PCA), a method of color component prediction based on independent component analysis (ICA) is proposed, which aims to search statistically independent not only non-correlated components. Time-consuming computing of Jacobian matrix in each iteration of the fast ICA algorithm influences its iteration performance. Modified fast ICA was advanced after analyzing the kernel iteration of the fast ICA algorithm, which reduces the degree of computing complication and improves the convergence rate. However, the convergence rate and resulted independent components of the modified fast ICA and fast ICA algorithms were dependent on initial random vector. The damped modified fast ICA was finally proposed by importing a dynamic dampness factor in the fast ICA algorithm before using modified fast ICA to gain independent components. The experimental results show that the novel method of color component prediction not only uncovers the real color components of the target image completely but reconstructs the spectra data set with a high colorimetric and spectral accuracy.