A New Method of Image Feature Extraction and Denoising Based on Independent Component Analysis

Sparse coding is a method for finding a representation of data in which each of the components of the representation is only rarely significantly active. Such a representation is closely related to independent component analysis (ICA), and has some neurophysiological plausibility. In this paper, we show how to choose the optimal sparse coding basis for denoising and how to apply an improved shrinkage operation on the components of sparse coding so as to reduce noise. Compared to the method of wavelet shrinkage, our method has the important benefit that the features are estimated directly from data. We also show a new approach of sliding window to improve the efficiency of sparse code shrinkage for realtime processing.