| Author |
 Topic  |
|
|
JannieT
Singapore
7 Posts |
 Posted - 02/18/2018 : 02:04:32 AM
|
Hi,
I have 2 Terahertz Time Domain Signal for fish samples, obtained from Terahertz Reflection Spectroscopy as seen below.
I would like to obtain a graph for the n of complex refractive index of the sample. And this is the original graph I obtained for n.
As the graph for n is very oscillatory and has huge y value range, therefore I thought the problem lies in the reference time domain signal. Therefore, I went to correct time domain signal until as such.
The corresponding change to n is as such. It has small y value variation. However, the number of oscillation increases. Seems like there is a lot of noise. And I performed curve fitting (as shown by the red line), but this does not look right too as n should be greater than 1.
Can I ask how to remove the noise for the signal such that I can perform a better curve fit to my signal?
Any help will be deeply appreciated. Thank you!
|
|
|
arstern
USA
237 Posts |
|
|
JannieT
Singapore
7 Posts |
Posted - 02/22/2018 : 6:43:30 PM
|
quote:
Originally posted by arstern
Hi,
You may want to try the smoothing feature:
https://www.originlab.com/doc/Tutorials/Smoothing
-Aviel
OriginLab
Hi Aviel,
Thanks for your reply.
Can I ask if there is a way to retain the true edge of the signal and size of the true edge?
As I have applied the methods in the link, they are either too out of shape, or edges may be gone, or may have additional edge coming out.
Below are some of the smoothed signals which are closer to the original signal.
Original Terahertz Time Domain Signal
Wavelet Denoise Signal
Smoothing by wavelet
Savitzky- Golay
 |
 |
|
|
Echo_Chu
China
Posts |
|
| |
Topic |
|
|
|
How to remove noise for signal or curve fit signal