摘要
利用Likas等处理图像恢复的变分法来讨论地震记录反演中的盲目反褶积问题.利用Kullback-leibler信息测度获得一个非常有用的统计函数(即变分函数),此函数的极值点就含有所要求的子波及反射系数序列,并且拟订出求此泛数极值点的一套算法.
In the inversion of the seismic,solving the problem on a blind deconvolution is pretty difficult.There are various methods to discuss this problem,such as the method of the cumulant moments etc.In this paper,the problem on a blind deconvolution in the seismic is discussed by using the functional method,by which Likas et al studied the problem on the restoration of an image.A statistical function(i.e.the variational function),the maximum point of which includes the wavelet and the coefficient sequence of the reflections,is obtained by using the Kullback-Liebler information measure,and the algorithm is devised for computing the the maximum value of the variational function.
引文
[1]J.A.Garcia,et al.Information theoretic measure for visualtarget distinctness[J].IEEE Trans on PAMI,2001,23(4):362-382.
[2]S.R.Kulkarmi,et al.Learning decision rules for patternclassification under afamily of probability measures[J].IEEETrans I.T.,1997,43:154-166.
[3]A.C.Likas,etc.A variational approach for Bayesian blindimage deconvolution[J].IEEE Trans on Signal Proc,2004,52:2222-2233.
[4]Lalear,G.D.Mixed-phase wavelet estimation using fourth-or-der cumulants[J].Geophysics,1993,58:1042-1049.
[5]K.Dogancay,etc.Blind detection of equalization errors incommunication systems[J].IEEE Trans I.T.,1997,43:469-482.