Detecting and changing the polarity of signals

Hi every one;

I have hundred of data files (signals) of equal length. I am interested in the highest peak (maximum) and location in each signal but want to have them with positive polarity. Now the problem is that, randomly, some signals have highest peak as positive polarity and some have negative. I want all of the signals with the positive polarity. Manually, I can just plot and if find negative, I multiple with -1, then find maximum and its location but my question is that can I do some thing that, it sees if highest value appears as negative, it converts the signal to positive. I will appreciate your help.

[Adam Danz]

Why not use the absolute value of your data?

[maxVal, maxIdx] = max(abs(data))

[Khan Engr]

Thanks Adam for looking into this.

Yes we can do that, but the thing is that, after finding the location I have to extract that highest pulse from the signal and I have to use a baseline value (because the pulses do not always coming to zero) for that which is for example around 0.002. Its so complicated to work with both positive and negative pulse extraction, infact I dont know how to do. I feel, its easy for me to have all the signals in positive and then to carry out further work.

[Adam Danz]

Could you share a plot of the data?

[Khan Engr]

i dont know how to implement it in Matlab, it can be something like

If

Max(data) is negative

then

data1=-data

else

data1=data

[Adam Danz]

I made some suggestions in the answers section below but I still can't picture your data. A plot would be helpful.

[Khan Engr]

Thanks,i will see it now and as you mentioned about plot, I am attaching a file having three such data files, you may please have a look in the and also a PDF that what I am looking for.

[Adam Danz]

I looked at the plot in the pdf file. If you have an estimate of the baseline (~-0.75 in these data) I'd go with my first option proposed in my answer. Subtract that from the signal so the new baseline is 0. Then you can just take max(abs()) + baseline.

Answer 1

[Adam Danz]

(Continuing from comments under the question) I think the best solution would be subtract the baseline value and then take the absolute value of your signal. Then you just need to find the max. I don't have your data but it would look something like this:

[maxVal, maxIdx] = max(abs(signal-baseline));

An alternative is to find the max and min, determine which has the greater magnitude, and then multiply the value by the sign of that number. If the sign is positive, the value will remain positive. If the sign is negative, multiplying two negatives will turn it positive. That would look something like this:

Assuming you have the peak value ( peakVal ) and its location ( peakIdx ),

peakVal * sign(data(peakIdx))

[Khan Engr]

Hi Adam, I am sorry, I was on travel and did not have opportunity to see this at all.

About ''An alternative is to find the max and min, determine which has the greater magnitude, and then multiply.....''

This could be good option more than the base line stuff I guess. for example

[minVal] = min(data1)

-0.01

maxval=max(data1)

0.005

then how should i proceed next, i mean writing script considering and if and else for positive and negative , as you suggested.

[Adam Danz]

The method below finds which element of the vector data1 has the largest magnitude (positive or negative) by using the absolute value and the n multiplies it by the sign of that number so negative peaks remain negative.

data1 = [-3     3    -2   -20     4    -1    -2     1    -2    -1]; 
[peak, peakIdx] = max(abs(data1)); 
peak = peak * sign(data1(peakIdx)); 
peak =
   -20

[Khan Engr]

OK Great... I have used it like;

 data1 = [-3     3    -2   20     4    -1    -2     1    -2    -1]; 
[peak, peakIdx] = max(abs(data1)); 
peak = peak * sign(data1(peakIdx)); 
>> if peak<0;
data2=-data1;
else
data2=data1;
end

Its working well for the purpose.

The problem is solved, thanks a lot Adam.