Blind Deconvolution of Noisy Seismic Trace Using Independent Component Analysis

From March 2004 to April 2005
Project Leader(s): Dr. S.S. Dlay, Dr. W.L. Woo
Sponsors: Foster Findlay

The ability to resolve thinly bedded strata in seismic data is determined by the characteristics of the seismic imaging system and in particular the transmitted seismic wavelet. The ability to recognise and characterise subtle stratigraphic changes is limited by the signal to noise characteristics of the seismic data. Improving resolution and signal to noise ratio are crucial to obtaining value from expensively acquired seismic data. Conventional techniques for enhancing resolution tend to reduce signal to noise ratio and techniques for removing noise lead to a loss in resolution. Independent Component Analysis (ICA) is a technique for separating mixed or convolved signals into statistically independent components with little or no a priori information. The hypothesis to be investigated in this project is that ICA will enable seismic traces to be separated into input wavelet, earth response signal and noise enabling significant improvements in seismic resolution to be achieved whilst simultaneously removing noise.


Professor Satnam Dlay
Professor of Signal Processing Analysis

Dr Wai Lok Woo
Director of Singapore Operations