How to keep main signal and suppress/remove other signal?

clc
clf
close all
clear all
%%
%import spectrum
load('I_d.mat');
load('depth.mat');

Fs = 200*1e3; %sampling rate: 200 kHz
time = 1/Fs:1/Fs:0.006; %time interval (unit:s)

figure(1)
plot(time,I_d);
title('original signal');
xlabel('time (s)');
ylabel('Amplitude');

signal = I_d;
window = hanning(length(signal));
signal = signal.* window;
FFT = abs(fftshift(fft(signal,2048)));
FFT = FFT/max(FFT);


figure(2)
plot(depth,log(FFT),'linewidth',1);
title('after doing FFT')
xlabel('Depth (µm)');
ylabel('Amplitude');
set(gca,'linewidth',1,'fontsize',15);
grid on


%% plot spectrum with the other method
Fs = 2001e3; %sampling rate
time = 1/Fs:1/Fs:0.006; %time interval (unit:s)
%Default window is hamming window
figure(3)
pwelch(signal,[],[],[],Fs); %[] length of window to be used

The main signal is at the 2000µm depth (figure).

How do I suppress other signal besides this main signal and maintain signal to noise ratio?

Answer 1

[Image Analyst]

That's not PSF (Point Spread Function). pwelch() computes PSD (Power Spectral Density). Basically you can fft the signal, then zero out all elements except those at or around 2000 and -2000, and then inverse transform.

[Image Analyst]

Not sure why your x axis has units of depth (space domain) instead of Hz (frequency domain). Can you explain?

[Vivian Yu]

Thanks for your reply. Acturally, the depth information depends on frequency because of Wiener-Khinchin theorem. The difference of optical pass length between reference arm and sample arm causes time difference. This concept is based on swept-source optical coherence tomography.