Algebraic Reconstruction Algorithm

function [x,rs,rho] = kaczmarz(A,b,sweeps,os) x = zeros(size(A,2),1); N = sqrt(length(x)); r=size(b,1); omega = 0.5; temp=x; rs = zeros(sweeps,1); rho = rs;

for k = 1:sweeps for j = os x = x + omega./(A(j,:)*A(j,:)')((b(j)-A(j,:)*x).*A(j,:))'; end rs(k)=norm(x); rho(k)=norm(b-A*x); disp(k); end;

This is the code i used for reconstructing a phantom. but through this i got an output with artifacts. I think i have to use convergence. kindly give me some ideas and help me to come out of this problem.

Answer 1

[Bharathi]

From net I got a point of convergence i.e b(j) should be defined as

    -

b = b + δb
where as δ is an noise factor.I don't know how to calculate

-

b