finding the max/peak of signal for a given vicinity of points without using matlab native functions

So, basically I'm trying to create a function that allows me to detec the peaks/maximums of a signal without using matlab native functions (basically max, min, findpeaks, etc --- because I can still use length, floor, ceil, etc).

And I've written a code that I thougth should do the trick, but it shows up as an error and I can't understand why.

This is what I have so far:

function [Picos] = detect_peaks(FINAL,k)
L = length(FINAL);
Picos = [];
if k/2==floor(k/2)
    par=1; % k é par
else 
    par=0; % k é ímpar
end 
if par==0
    a = (k-1)/2; 
else
    a=k/2;
end
for i = 1:L
    if i<(a)
        for q = 1:k
            if FINAL(i) > FINAL(q)
                max = FINAL(i);
            else
                max = FINAL(q);
            end
            
        end
    elseif i>(L-a)
        for q = (L - k):L
            if FINAL(i) > FINAL(q)
                max = FINAL(i);
            else
                max = FINAL(q);
            end
            
        end
    else
        for q = (i-(ceil(a))):(i + (floor(a)))
            if FINAL(i) > FINAL(q)
                max = FINAL(i);
            else
                max = FINAL(q);
            end
        end
        Picos(i) = max;
    end
end

--------

All of this considering that k is the vecinity (in which the the value associated whith the indice in question is included ) and FINAL is a signal that looks something like the image attached.

I would really apreciate it if someone could help me.

[Dyuman Joshi]

It's not clear what you want to do. You want to find the maximum in the range [-k k] of all the indices? And what other functions are not to be used?

What is the use/purpose of variable "par"?

function [Picos] = detect_peaks(FINAL,k)
L = length(FINAL);
Picos = zeros(1,L);
if k/2==floor(k/2)
    par=1; % k é par
    a=k/2;
else 
    par=0; % k é ímpar
    a=(k-1)/2;
end 
for i = 1:L
   if i<(a)
       for q = 1:k
           if FINAL(i) > FINAL(q)
               maxval = FINAL(i);
           else
               maxval = FINAL(q);
           end
           
       end
   elseif i>(L-a)
       for q = (L - k):L
           if FINAL(i) > FINAL(q)
               maxval = FINAL(i);
           else
               maxval = FINAL(q);
           end
          
       end
   else
       for q = (i-(ceil(a))):(i + (floor(a)))
           if FINAL(i) > FINAL(q)
               maxval = FINAL(i);
           else
               maxval = FINAL(q);
           end
       end
     Picos(i) = maxval;
   end
end

[Inês Silva]

I'm sorry I wasn't clear. The par variable is written in my mother thongue , which is my it isn't selfexplanatory to english speakers, but basically it just checks if the vicinity (k) is an odd or an even number. The way the vicinity works is that if the vicinity is for instance 5 the program is gonna check the maximum in the interval of (indice - 2) to (indice + 2) and if the vicinity if for instance 6 then by defaul the interval is going to go from (indice-3) to (indice + 2) and so on. This is to the fact that the indice itself counts as a member of the vicinity .

The objective is to obtain the peaks in the signal ( what is marked in red in the image bellow )

The only thing is that the code needs to work based on the vicinity method I explained above.

The functions I'm not allowed to use are the following:

mod, medfilt, spit, sort, max, min, rem, findpeaks, global, mean

Thank you once again and I hope this helps clear things up a bit.

[Dyuman Joshi]

I understood about the vicinity, but what is the use of par variable in the rest of your code?

movmax is fits nicely with what you want to do.

Also, in finding peaks, how to discard the small peaks that occur in the graph (in the 50-100 range) - Do you have any info regagrding that? Or we have to take into account in the code itself?

[Inês Silva]

You are right the par variable wasn't needed.

Unfortunately I also can use movmax.

And to eliminate the smaller peaks I use the limiar variable.

function [Picos] = detect_peaks(FINAL,k,limiar)

L = length(FINAL);
Picos = [];

a = (k-1)/2; 


for i = 1:L
   if i<(ceil(a))
       for q = 1:k
           if FINAL(i) > FINAL(q)
               max = FINAL(i);
           else
               max = FINAL(q);
           end
           
       end
   elseif i>(L-(floor(a)))
       for q = (L - k):L
           if FINAL(i) > FINAL(q)
               max = FINAL(i);
           else
               max = FINAL(q);
           end
          
       end
   else
       for q = (i-(ceil(a))):(i + (floor(a)))
           if FINAL(i) > FINAL(q)
               max = FINAL(i);
           else
               max = FINAL(q);
           end
       end
   end
   Picos(i) = max;
end
 
for k = 1:(length(Picos))
    if picos(k) < limiar
          picos(k) = [];
    end
end

[Dyuman Joshi]

Can you use islocalmax ?

[Inês Silva]

unfortunately no.

I can only use the general functions such as clc clear close, functions for the entry or exit of data such as input xlsread xlswrite..., functions related to the creation of matrixes such as eye , ones, isempty .... as well as length and size , functions related to the conversion of values such as str2num double, uint8.... and functions related to graphic representation such as plot, title , scatter...

[Inês Silva]

basically what I need is a way to obtain the same results I woulg get if I used findpeaks without using findpeaks and basically only using repeticion cycles ( for and if and switch )

[Inês Silva]

I've given another try at the code and came up with this:

function [Picos] = detect_peaks(FINAL,k,limiar)

L = length(FINAL);
Picos = [];

a = (k-1)/2; 


for i = 1:k:L
   if i<(ceil(a))
       maximo = 0;
       for q = 1:k
           if FINAL(q) > maximo
                 maximo = FINAL(q);
                 Picos(i) = [maximo,q];
           end   
       end
   elseif i>(L-(floor(a)))
       maximo = 0;
       for q = (L - k):L
           if FINAL(q) > maximo
               maximo = FINAL(q);
               Picos(i) = [maximo,q];
           end
          
       end
   else
       maximo = 0;
       for q = (i-(ceil(a))):(i + (floor(a)))
           if FINAL(q) > maximo
               maximo = FINAL(q);
               Picos(i) = [maximo,q];
           end
       end
   end
  
end

The problem is that this error shows up:

Unable to perform assignment because the indices on the left side are not compatible with the size

of the right side.

Error in untitled (line 26)

Picos(i) = [maximo,q];

But if I don't put the 'i' in Picos(i) then only the last maximum / peak is registred in the matrix.

[Inês Silva]

And this is the last update on the code:

But there is still an error of some sort because the values don't match what I can see in the graphic

function [Picos] = detect_peaks(FINAL,k,limiar)

L = length(FINAL);
Picos = [];

a = (k-1)/2; 

t = 1;
for i = 1:k:L
   if i<(ceil(a))
       for q = 1:k
           maximo = 0;
           if FINAL(q) > maximo
               maximo = FINAL(q);
               M = q;
           end
       end
       Picos(t,1) = maximo;
       Picos(t,2) = M;
       t = t + 1;

   elseif i>(L-(floor(a)))
       for q = (L - k):L
           maximo = 0;
           if FINAL(q) > maximo
               maximo = FINAL(q);
               M = q;
           end
       end
       Picos(t,1) = maximo;
       Picos(t,2) = M;
       t = t + 1;
   else
       for q = (i-(ceil(a))):(i + (floor(a)))
           maximo = 0;
           if FINAL(q) > maximo
               maximo = FINAL(q);
               M = q;
           end
       end
       Picos(t,1) = maximo;
       Picos(t,2) = M;
       t = t + 1;
   end
  
end
m = 1;
Peaks = [];
for p = 1:t
    if Picos(p,1) > limiar
        Peaks(m,1) = Picos(p,1);
        Peaks(m,2) = Picos(m,2);
        m = m + 1;
    end
end