Looping for find peak of signal

Hai...i have a problem to save looping data. For one row signal, my matlab is OK, but if my input more than one row signals it's a trouble. I want to find peak of signal but the peaks of every signal is diffent...please help me

 clear erp 
%Data
 Ypeak = D3_Kiri_Bior35
 %Ypeak = Ypeak(2:2,:)

 [M1 M2] = size (Ypeak)

 for M3 = 1:M1

 Xpeak = Ypeak (M3:M3, 1 : M2)  
 %Xpeak = Ypeak
 V     = std(Xpeak)
 M     = mean(Xpeak)
 S     = 2*(M+V)

 [maxmin] = find(Xpeak > S);
 [minmax] = find(Xpeak <-S);

 %Find ERP
 Apeak     = maxmin(1,:);
 Bpeak     = minmax(1,:);
 Bpeak     = Bpeak'
 [ap1,bp1] = size(Xpeak)
 [ap2,bp2] = size(Bpeak)

 kurang = Bpeak(:,1:1)-30;
 tambah = Bpeak(:,1:1)+30;

%Loop [Edit SCd: code formatting]
for ipeak = 1:ap2   
     if   kurang(ipeak) <= 0
          kurang(ipeak) = Bpeak(ipeak,1:1)-(Bpeak(ipeak,1:1)-1) 
     else kurang(ipeak) > 0
          kurang(ipeak) = Bpeak(ipeak,1:1)-30
     end
end

 for kpeak = 1:ap2      
       if   tambah(kpeak) >= M2    
            tambah(kpeak) = Bpeak(kpeak,1:1)+ (bp1- Bpeak(kpeak,1:1))     
       else tambah(kpeak) < M2
            tambah(kpeak) = Bpeak(kpeak,1:1)+30

       end
 end

 for iipeak = 1:ap2   
    erp=Ypeak(1:1,kurang(iipeak:iipeak,1:1):tambah(iipeak:iipeak,1:1))
 end

 end

Answer 1

[Sean de Wolski]

why can't you just use max or findpeaks?

More per comments

%data 1 valley, two peaks
x = rand(1,100)*10;
x(20) = -3;
x([52 72]) = 7*pi;

the_mean = mean(x);
stddev  = std(x);
low_lim = the_mean - 2*stddev; %lower limit
up_lim  = the_mean + 2*stddev;

idx_up = find(x>up_lim);  %locations
idx_low = find(x<low_lim);

xup_parts = x(min(max(bsxfun(@plus,idx_up(:),-30:30),1),length(x)));  %indices of parts
xlow_parts = x(min(max(bsxfun(@plus,idx_low(:),-30:30),1),length(x)));

[Endro Yulianto]

ok...but how to use that for more than 2 row signal?

[Endro Yulianto]

i mean if i have 2 signal that in every signal maybe have 2 until 3 peak

x = rand(2,100)*10;

haven't i use "for loop"?

That my problem

[Sean de Wolski]

Yeah, just loop over every row:

for ii = 1:size(x,1)

do_stuff_with(x(ii,:))

end

[Endro Yulianto]

%data 1 valley, two peaks

x = rand(2,100)*10;

x(:,20) = -3;

x(:,[52 72]) = 7*pi;

for ii = 1:size(x,1)

the_mean = mean(x(ii,:));

stddev = std(x(ii,:));

low_lim = the_mean - 2*stddev; %lower limit

up_lim = the_mean + 2*stddev;

idx_up = find(x(ii,:)>up_lim); %locations

idx_low = find(x(ii,:)<low_lim);

xup_parts = x(min(max(bsxfun(@plus,idx_up(:),-30:30),1),length(x))); %indices of parts

xlow_parts = x(min(max(bsxfun(@plus,idx_low(:),-30:30),1),length(x)));

end

xup_parts

xlow_parts

[Endro Yulianto]

It have to be 2 xup_parts and 1 xlow_parts from first signal and 2 xup_parts and 1 xlow_parts from second signal. But i only get 2 xup_parts and 1 xlow_parts

[Endro Yulianto]

why always only data from last loop that saved?

[Sean de Wolski]

because you overwrite it on each iteration. You need to save to the current location. The easiest way would just to be

xup_parts = [xup_parts;new_x_up_parts];

Add the old one to the end.

[Endro Yulianto]

??? Undefined function or variable "new_x_up_parts".

Error in ==> cobapeak1 at 24

xup_parts = [xup_parts;new_x_up_parts];

i'm sorry i'm beginner in matlab

[Sean de Wolski]

that's my way of trying to hint that the new computation on each iteration of the for-loop. It wasn't working code.

[Endro Yulianto]

that my problem...i don't know how to save each iteration to cuurent location so i save data from all looping

[Endro Yulianto]

please help me

[Endro Yulianto]

I do it for 2 week but i don't know how to save looping in current location

Answer 2

[Endro Yulianto]

I try to find peak of signal that amplitudo of the peak is above and under mean (+/-) 2*standart deviation and from that peak i segmenteg that signal (+/-) 30 data with peak is the center. The problem is when i try to find more than 2 row signal from looping, only last data is saved. Please help me

[Sean de Wolski]

Calculate the std multiply by two and add/subtract from mean. Then use logical indexing to find what's outside that? I'm sorry, I'm having trouble understanding you.

[Endro Yulianto]

For example : Mean= 2, std = 1. I try to find peak that have amplitudo above +(2 + 2(1)) and under -(+(2 + 2(1)). If i find peak at 100, i'll segmented data at (100 - 30) until (100 + 30) = 70 : 130.

[Image Analyst]

Can you give us some MATLAB code to generate some sample data, that we can then use to create a demo for you?

[Endro Yulianto]

x = rand(2,100)*10;

x(:,20) = -3;

x(:,[52 72]) = 7*pi;