Hi all,
I'll have to filter Raman spectra and I would like to do so in R. I'm
trying
to figure out how. Maybe one of you could help me ...
This is what I have to do:
General:
separate the real spectrum (the set of peaks) from two
contaminations: a drifting background and (white) noise
Constraints (which should help the task):
A - the spectrum is positive
B - the spectrum is long (covers about 1000 wave numbers)
C - the background changes slowly (on the order of 10 equivalent
degrees of freedom over the 1000 wave numbers).
D - the spectrum changes fast (a peak comprises 10-20 wave-numbers)
E - the noise changes very fast (can be assumed uncorrelated)
F - sometimes, repeat measurements are available on the same (or
essentially the same) analyte: such repeats are
supposed to share the same spectrum but may have different
backgrounds (and the noise is certainly different)
Ideas: (on the lines of smoothing)
Use a positive smoother to model signal (maybe a finite support
kernel of appropriate bandwidth)
Use a smoothing spline with low degree of freedom (10 or so) for the
background
Use a generalized additive model type setup for fitting (with
Gaussian error)
Assistance would be greatly appreciated.
Chris
Christian Ritter, Statistical Services, Belgian Shell Chemicals
Tel. +32 10 477349, fax +32 10 477219, email christian.c.ritter at opc.shell
com
-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-help mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !) To: r-help-request at
stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._
