Bind Deconvolution Based On Independent Component Analysis

Deconvolution is the key of seismic data processing to improve the resolution of seismic data. The traditional methods of seismic deconvolution are always based on the assumpsition of Guassality and whiten noise to reflection coefficient and minimum phase to the seismic wavelet. Generally, these methods can obtain a good effect when they are used in the practical application with the above-mensioned assumpsition. However, it can’t guarantee the accuracy of these assumpsitions when they are used in the seismic data processing. Blind deconvolution can achieve the seismic wavelet and reflection coefficient simultaneously without the assumpsition of Guassality and whiten noise to reflection coefficient and minimum phase to the seismic wavelet. The results obtained from the blind deconvolution are more suitable to the real condition. So it has a realistic significance and developing prospect to do research on the blind deconvolution. The research is financially supported by the National High Technology Research and Development Program (863 program) of China (Grant No. 2006AA06A108).Blind deconvolution drives from the blind singnal separation and has been widely appreciated in many fields, especially in the speech enhancement, data mining, medical signal analysis and processing, image recognition and wireless communication. It also showed the initial application of noise removing, multiple wave attenuation and seismic deconvolution in geophysical field, and the practical use becomes more and more important development trend.Independent component analysis is a new multi-dimensional signal processing method based on high-level statistics, which is able to achieve the separation of source signals under the absence of priori information. The observed signal will be divided into a number of independent signal components, thus helping to enhance and analyze the signals. Considering its wide and attractive applications, many researchers have studied independent component analysis in the past thirty years so that independent component analysis teehnique has been developed very much. However, independent component analysis is still staying at the developing stage and the investigation of its theoy and application should be ehanced and improved further.The paper studies the theory and algorithm of blind deconvolution based on independent component analysis on the researches of predecessors. On the basis of the study, this dissertation achieves the following breakthrough and innovations:1. Introducing the basic principles and methods of blind deconvolution systemly.2. Introducing the basic theory, models, optimization algorithms, optimization criterion of independent component analysis and discusses the main criterion for independent component analysis which includes information maxim- ization, the maximum likelihood estimation, mutual information minimization and non-Gaussian measure. The main algorithm includes: Joint approximate diagonal- ization method (JADE), fast fixed-point iteration method, information maximization.3. Neglecting noise, the seismic blind deconvolution has been achieved in both time and homomorphic domain. The Gaussian white noise is added to the blind deconvolution based on independent component analysis in time domain which drives a new algorithm which is called noise blind deconvolution based on independent component analysis.4. Some research is also been done on the multichannel blind deconvolution based on the constrained multiuser kurtosis optimization criterion, which results in a new denoising algorithm based on the constrained multiuser kurtosis (MUK) optimization criterion. The independent component analysis is also been involved in which drives a new algorithm which is called the multichannel blind deconvolution based on independent component analysis.5. Using noise blind deconvolution based on independent component analysis to achieve the drill signal to substitute the traditional drill deconvolution to obtain the high resolution of the seismic exploration while drilling (SWD) single shot record. Using the multichannel blind deconvolution based on independent component analysis to achieve the SWD single shot record directly.The model and real seismic data mumerical examples all shows that the noise blind deconvolution based on independent component analysis and the multichannel blind deconvolution based on independent component analysis can improve the resolution the SWD single shot record and has a good application prospects.