when the SNR is low, how to improve peak detect result

[frogliang]

There are some points that missed by the peak detector, but if i lower the threshold or width, the higher signal counted double peaks. i guess that's due to the low SNR. cause in the picture below, the peak detector is well function. so how should i improve the peak detection?

[Bob_Schor]

It's pretty difficult to make suggestions from a picture of the output of your program.  We would need the Program (as code that we could run, observe, and "attempt to improve") and the data set that you used to produce the pictures you showed.  It is impossible to tell if there is a problem, much less where it might be or to suggest how to correct it.

Bob Schor

[billko]

How can I improve the CO2 levels in my car's output?

You get the picture (pun intended)?

[Bob_Schor]

Zip -- right over my head (but a very nice picture ...).

Bob Schor

[frogliang]

sorry, I thought that is a common topic of signal processing. I guess there are solutions like some kind of filter. but the noises contain various frequency, i tried filter, but not good.

Thanks

[altenbach]

The noise is quite constant and your "peaks" seem extremely sharp and easily distinguishable from the noise with a simple threshold detector. Any "filtering" would broaden them into oblivion, making detection more difficult. What else do you know about the peak shape (gaussian, lorentzian, rectangular, etc.) Do the peaks vary in area? What is their origin? Do you have the wavelet toolkit?

In any case, the more you know about the signal, the easier it will be to separate the interesting parts from noise. Tell us as much as you can!

[altenbach]

Also, your VI is not useful to us. Attach relevant data instead.