Generate Sparsity basis matrix

In this demo :

function  [xx, Psi, al] = GenSyn(n,k,nist,mode,Psi) 

The Mandatory arguments are :

% n = signal dimensionality

% k = sparsity level % nist = amount of generated instances

and the Outputs are :

% xx = matrix containing the generated instances

% Psi = corresponding sparsity basis

% al = matrix containing the sparse representation of xx

% Written by Mauro Mangia and Riccardo Rovatti.
clc;
clear ;
close all;
cdir = pwd;

if strcmp(cdir(end-3:end),'demo')
    cd ..
    addpath(genpath(pwd))
    cd demo
end

GetSNRdB = @(Data, DataX) 20*log10(norm(Data)/norm(Data-DataX));
opts = spgSetParms('verbosity',0);  % Turn off the SPGL1 log output
verbosity=2; %Turn on the BRak log output

%% setting
n=384;
k=5;
mode='LP'; %(LP for low-pass profile and BP for band-pass profile)
%% correlation matrix and dataset for the considered setting
% input signals
nw=3; %amount of signal instances used as dataset
[sig, F]=GenSyn(n,k,nw,mode);
%% TESTs
RND = zeros(size(sig));
for is = 1:nw
    x=sig(:,is);
    %% standard CS-Antipodal
    % sensing matrix
    phi = ceil(rand(M,n)-1+p);
    %measurement vector
    y=phi*x;
    %signal reconstruction
    RND(:,is) = F*spgl1(phi*F, y,0,1e-5,[],opts);
end

I already have a sparse signal, I do not need to generate one. What I need is just to know how to find the matrix psi, which is the corresponding sparsity basis, using Matlab?