kl.m

著名的seismiclab的代码 是地震学研究人员必备的工具

function [D_out,s] = kl(D_in,P);  
%KL: Karhunen Loeve filtering of seismic data.
%    The Karhunen Loeve transform can be used to enhance
%    lateral coherence of seismic events, this transform 
%    works quite well with NMO corrected CMP gathers, or
%    common offset sections.
%
%  [D_out,s] = kl(D_in,P)
%
%  IN   D_in:  data (the data matrix)
%       P:     number of eigenvectors to used by the
%              KL filter
%
%  OUT  D_out: output data after kl filtering
%       s:     the eigen-values of the covariance
%              matrix in descending order
%
%  Example : Filter a cdp; only 5 eigenvectors
%            are kept
%
%            [D_in,H]=readsegy('cdp.linux.su');
%            [D_out,s]=kl(D_in,5); 
%            wigb(D_out);
%
%  SeismicLab
%  Version 1
%
%  written by M.D.Sacchi, last modified March 4, 1998.
%  [e-mail] 
% 
%  Copyright (C) 1999 Seismic Processing and Imaging Group 
%                     Department of Physics
%                     The University of Alberta 
%
%

[nt,nh]=size(D_in);
D_out=zeros(nt,nh);

% Get data covariance matrix

R = D_in'*D_in;

[U,S]=eigs(R,'LM');

U=U(:,1:P);
D_out=(U*U'*D_in')';
s = diag(S);

return