Gaudier,
there is no easy solution as one needs some knowledge about the
properties of the signal in order to eliminate just noise and not
important signal properties.
Reviewing your chart there are two gaps or valleys in the signal. Do
you count 'em as noise or as relevant? If they are relevant, what is
the relevant property? Width, peak value, slope ...
You can always start with a moving average over 3, 5, 7 or so points.
Or use the median out of 3, 5, 7 or so points. But both these methods
harm the peak (local extreme value) of that gaps. If you are interested
in these try to smooth your data with a spline function. And finally,
if you are 'just' interested in the peak value, perform a square fit
over a symmetric number of points around the actual local
maximum. A local max is reached where the slope of the sgnal is
zero. So differentiate the formual and set it to zero. Just some math...
Greetings from Germany!<br>-- <br>Uwe
