further noise reduction techniques ontop of Savitzky-Golay FIR filter

Ian_Brooks

Hi PF Designers,

Wasn't too sure if this would belong in the Electrical engineering forum but here goes.

I'm designing a prototype for my thesis that will read in sensory data wirelessly to a matlab script. This sensory data is riddled with noise, however important data is contained in the High frequency components of the signal as well.

Hence I was naturally going to use a Savitzky-Golay FIR filter in matlab. However I was wondering if I could perform further noise reduction techniques on top of this and still preserve the high frequency component that I saved.

Thoughts, dilemmas, discussion, tid bits of help?

Erez_L

Do you have any theoretical model?

If so there are many thechniques (e.g. kalman filter).
http://en.wikipedia.org/wiki/Kalman_filter

If not, instead of using S-golay, you can try building a specific LP filter and play with the threshold frequency

marcusl

If you know the form of your input data, then a Wiener filter will give you the theoretically optimum SNR at the output. For white Gaussian noise you can implement the optimal filter as a matched filter.