CODEWITHSUNDEEP
matlabsignal-processingvectorization
user2041376
user2041376

Reputation:

best way to count pulses in 1-D waveform in Matlab

I have a signal composed of square pulses (+ some noise), here's a tiny part of it:

enter image description here

I look for an efficient and robust way to count how many pulses I have.

Here's what I've done so far:

The amplitude is a bit noisy but SNR is great do I can threshold: 

data = data>1;

the length of each pulse can be noisy so I ignore it and use diff, to obtain the derivatives (+ and -), find how many non-zero elements there are, and divide by 2 (since there are 2 derivative peaks per pulse) . 

dd=diff(data);

num_of_pulses=length(find(diff(dd)))/2

Is that the best way to do that? I was told not to use diff because it can be too noisy...

Upvotes: 2

Views: 4028

Answers (1)

shoelzer
shoelzer

Reputation: 10708

Based on your description of the data, I think this will work.

numberOfPulses = nnz(diff(data > 1) > 0)

You can reliably find pulse samples using data > 1, then use diff() > 0 find the transitions from no pulse to pulse, and finally nnz() to count them.

Upvotes: 3

Related Questions