PsC.m

function [PsC_s,best_lag] = PsC(template,sig,lag)
%% Psuedo_Correlation
% Computes the Pseudo Correlation , a finer approach than normal
% correlation for template matching. Please review the paper below in order
% to see why it is much more accurate for the pattern recognition
%% Input
% template : template to be searched in the signal
% sig: the signal that we are searching the template in
% lag : does PsC for the lag between -lag : lag, it should be in samples,
% for example half the length of the input signal
%% Output :
% PsC_score : maximum score at best lag
%% Algorithm is based on the following paper :
% H. Sedghamiz and Daniele Santonocito,'Unsupervised Detection and
% Classification of Motor Unit Action Potentials in Intramuscular 
% Electromyography Signals', The 5th IEEE International Conference on
% E-Health and Bioengineering - EHB 2015, At Iasi-Romania.
%% Author: 
% Hooman Sedghamiz
% June 2015, Linkoping University
% Please cite the paper if any of the methods were helpfull

%% Script Begins here, Do not Change
%%
if nargin < 3
    lag = round(length(sig)/2);
end


template = template(:);
sig = sig(:);

m = length(template);
n = length(sig);

 if m > n
   error('Length of Template should be equal or smaller than pattern'); 
 end

 sig = [zeros(lag,1);sig;zeros(2*lag,1)];
 p4 = zeros(1,m);
 normaliz = zeros(1,m);
 PsC_score = zeros(1,n);
 
 for k = 0  : 2*lag
     for i = 1 : m
         p1 = (template(i)*sig(k+i));
         p2 = abs(template(i) - sig(k+i));
         p3 = max(abs(template(i)),abs(sig(k+i)));
         p4(i) = (p1 - p2*p3);
         normaliz(i) = p3^2;
     end
   PsC_score(k+1) = max(sum(p4)/sum(normaliz),0);
 end

 [PsC_s,best_lag] = max(PsC_score);
 best_lag = best_lag - (lag + 1);
end