A Blind Separation Method of Noised Image Based on Neural Network Nonlinear Filtering and Independent Component Analysis

The main objective of this work is to develop a new method for the blind separation of the noised image. A nonlinear neural network and independent component analysis (ICA) algorithm are combined. The neural network filter is used to remove the noise and ICA algorithm is used for the blind separation of the mixed image. But the effect of pre-filter is different from the post-filter. By comparing the experimental results, pre-filter is proved to be more effective. The research work is helpful for the blind source separation of the multidimensional signal.