How can I find cross-correlation between two signals?

Hello. :)

I heard that there are a lot of geniuses here. help me :(

I have a couple of question about my algorithm.

I want find similarity between two signal.

I generate two simple signal(temp, temp2)

I want find a similar signal between "red" signal and "blue" signal

So, I used xcorr fuction in matlab.

I used xcorr(red,blue) command in matlab.

Because red signal shape is similar to "1st Peak" at blue signal in image(attached image)

So the algorithm shoud find have found "1st peak" .

But my source was find always "2 peak" at blue line in image(attached image).

I was worried enough... But I can not find solution!

The problem is I think "2 peak" value amplutude is bigger than "1peak".

I think the result of convoultion(cross-correlation) value is big at 2peak....!

what should i do?

*I want find that where is the most similar signal. like X-axis value*

(I wnat to find "1 peak in blue lin")

The red signal time shifted just 10 times. not 110 times....

*The answer is 10!! :(*

Should I use Threshold? is it good algorithm? This case is very simple but if I solve the this problem, I'll apply this algorithm to complex signal.

here is my code

I just ask idea(like think, algorithm...) Not overall source code rivision.

2013.10.24 (10:36)

I tried command "corrcoeff".

But the R value is same between '1 peak' and '2 peak' as "1" value.

How can I classifiate these signl?

Have a good day!!

#####generate signal########

t3 = 1:0.01:1.5;
scanline4 = sin(2*pi*t3);
temp = [scanline4];
temp2 = [[zeros(1,10) scanline4 2*scanline4]];
windo = 50;
repe =1;
sector = 1;
count = 1;

for targetPt = 1:1 : length(temp)
    if (targetPt >= length(temp)-windo), break,end ;
    maxcorr = -10;
    for i = 1:repe:length(temp2)
        if (i >= length(temp2)-windo), break,end ;
        [correlation,lags] = xcorr(temp(targetPt:targetPt+windo),temp2(i:i+windo));
        c = xcorr(temp(targetPt:targetPt+windo),temp2(i:i+windo));        
        maxV(count)= max(correlation);
        if maxcorr < maxV(count)
            maxcorr = maxV(count);
            index = i
        end
        count = count +1;
    end
    delaymatrix(sector,1) =  index-targetPt;
    sector = sector+1;
end;

[Ramesh Bala]

Strange that no reply here !!

This is the syntex for cross correlation

crosscorr(y1,y2)

crosscorr(y1,y2,Name,Value)

xcf = crosscorr(___)

[xcf,lags,bounds] = crosscorr(___)

crosscorr(ax,___)

[xcf,lags,bounds,h] = crosscorr(___)