OMP算法的matlab实现

function [x,res] = omp(b,A,k)% compute k-sparse approximation to b with matrix A using Matching pursuit[m,N] = size(A);x = zeros(N,1);res = b;support = []; A_get=zeros(m,k);for i=1:kcorr = (A')*res;[c,n] = max(abs(corr)); support(end+1) = n;A_get(:,i)=A(:,n);A(:,n)=zeros(m,1);%x(support) = pinv(A(:,support))*bx(support)=(A_get(:,1:i)'*A_get(:,1:i))^(-1)*A_get(:,1:i)'*b;%res = res - A(:,support)*x(support)res=b-A_get(:,1:i)*x(support);%相乘为向量的正交投影end