Independent component analysis and its testing application on seismic signal processing

ICA is a novel statistical method developed recently, which is used to find a representation of the Non Gaussian multivariate data, so that each components of the vector are independent statistically, or as independent as possible. In many applications, the transformation aims to capture the basic structure in data, including features abstraction and signals separation. In this paper, we present the principal theory and fast algorithms, at the same time, realizing the FastICA and its slightly updating. On the base of analyzing the features of seismic signals, we do preliminary studies and try to apply ICA on seismic signal processing. Our works show the good perspective of ICA application on seismic signal processing.