How to remove negative peaks from a signal

How to remove negative peaks from a signal

[KALYAN ACHARJYA]

You can do it in multiples ways, Share the details, code too (if possible)?

[Image Analyst]

Define negative peaks. Is it everything between positive peaks? What does "remove" mean to you? Set to zero? Draw a line between the tips of the neighboring positive peaks? Delete the elements?

First read this link, then attach your data in a .mat file, and attach a screenshot of your data plotted plus an indication of what you'd like the signal to look like after the negative peaks are removed.

[sandhya sandhya]

I want to detrend the signal first and then i want to remove all the negative peaks from the signal.

Answer 1

[Image Analyst]

sandhya, it looks like you didn't even read the comments above since there is no plot of what you are starting with, no plot of what you'd want as an output, and no explanation of what defines a negative peak (valley).

I wrote a program to smooth the data and remove valleys so that the final signal is just lines between the peaks of the noise-reduced signal.

% Custom program for sandhya to remove valleys from a signal.
clc;    % Clear the command window.
clearvars;
close all;  % Close all figures (except those of imtool.)
workspace;  % Make sure the workspace panel is showing.
format long g;
format compact;
fontSize = 20;

s = load('s4.mat')
val = s.val
subplot(2, 1, 1);
plot(val, 'b-', 'LineWidth', 2);
grid on;
xlabel('index', 'FontSize', fontSize);
ylabel('val', 'FontSize', fontSize);
caption = sprintf('val vs. index for %d elements', length(val));
title(caption, 'FontSize', fontSize);

% Signal is pretty noisy.  Smmoth it with a Savitzky-Golay filter (moving quadratic).
filterWindowWidth = 91;
smoothedSignal = sgolayfilt(val, 2, filterWindowWidth);
subplot(2, 1, 2);
plot(smoothedSignal, 'b-', 'LineWidth', 2);
grid on;
xlabel('index', 'FontSize', fontSize);
ylabel('Smoothed Signal', 'FontSize', fontSize);
caption = sprintf('Smoothed Signal index for %d elements', length(val));
title(caption, 'FontSize', fontSize);

%------------------------------------------------------------------------------
% Set up figure properties:
% Enlarge figure to full screen.
set(gcf, 'Units', 'Normalized', 'OuterPosition', [0, 0.04, 1, 0.96]);
% Get rid of tool bar and pulldown menus that are along top of figure.
% set(gcf, 'Toolbar', 'none', 'Menu', 'none');
% Give a name to the title bar.
set(gcf, 'Name', 'Demo by ImageAnalyst', 'NumberTitle', 'Off')

% Find peaks in the smoothed signal - requires Signal processing Toolbox.
[peakValues, peakIndexes] = findpeaks(smoothedSignal,'MinPeakProminence', 3, 'Annotate', 'extents');  % Use default parameters
% Tack on first and last points.
peakIndexes = [1, peakIndexes, length(val)];
peakValues = [1, peakValues, val(end)];
% Plot results.
figure;
subplot(2, 1, 1);
plot(smoothedSignal, 'b-', 'LineWidth', 2);
hold on;
plot(peakIndexes, peakValues, 'rv', 'LineWidth', 2, 'MarkerSize', 10);
plot(peakIndexes, peakValues, 'r-', 'LineWidth', 1);
grid on;
xlabel('index', 'FontSize', fontSize);
ylabel('val', 'FontSize', fontSize);
caption = sprintf('Location of %d peaks', length(peakIndexes));
title(caption, 'FontSize', fontSize);

% Interpolate between peaks
allIndexes = 1 : length(val);
yInterp = interp1(peakIndexes, val(peakIndexes), allIndexes);
subplot(2, 1, 2);
plot(val, 'b-', 'LineWidth', 1);
hold on;
plot(allIndexes, yInterp, 'r-', 'LineWidth', 2);
grid on;
xlabel('index', 'FontSize', fontSize);
ylabel('fitted vals', 'FontSize', fontSize);
title('Final Signal in red with negative peaks (valleys) removed', 'FontSize', fontSize);
legend('Original Signal', 'Final, valleys-removed signal');

%------------------------------------------------------------------------------
% Set up figure properties:
% Enlarge figure to full screen.
set(gcf, 'Units', 'Normalized', 'OuterPosition', [0, 0.04, 1, 0.96]);
% Get rid of tool bar and pulldown menus that are along top of figure.
% set(gcf, 'Toolbar', 'none', 'Menu', 'none');
% Give a name to the title bar.
set(gcf, 'Name', 'Demo by ImageAnalyst', 'NumberTitle', 'Off')

I hope it's what you want. Since you didn't provide specifics, I guess you should be happy with anything that we do. I think I put more time into your problem than you did, so I hope my time was not wasted.

If it's not what you want, then re-read the comments above (under your original question at the top) and fix/improve your question.

[sandhya sandhya]

Thank you sir for your valuable reply. I want to remove all the negative peaks, because I am assuming that those are all disturbances or artifacts in my signal. So,there is no problem if you completely set those negative peaks to zero. Here I am attaching my file which I removed noise in the signal and also detrended the same signal.

[Image Analyst]

I inserted your picture for you so that we can see it without saving the .fig file to disk, then switching to MATLAB, then opening up the fig file in MATLAB. I saved it as a PNG file then inserted it into your comment above.

Anyway, I can't see what is different in this signal from the original signal. I also don't know what you define as a negative peak. Is it everything between positive peaks? If so, setting negative peaks to zero would just zero out the entire signal except for the very tips of the positive peaks, and that doesn't seem any good. Is it the part of your signal that is negative (has a value below zero)?

Also when you detrend it, over what span do you want to determine the trend? Ten elements? A hundred elements?

[sandhya sandhya]

The difference between original signal and the signal which i posted in previous comment window is,the original signal is noisy so i applied wavelet denoising technique to remove the noise in the signal.Here iam attaching my file so we can see the difference between the two signals.

[Image Analyst]

OK. I'll continue to wait for you to answer the other questions in my last comment. I'm in no rush.

[sandhya sandhya]

Remove Trends from Data I want to remove trends in the complete signal.so, I applied the technique (remove trends from data).

[Image Analyst]

That link doesn't work. Also I'm waiting to see an answer for every one of my sentences that had a question mark at the end of them. Otherwise I don't know that my answer would be different than the first answer for you, which I spent a lot of time on. I just don't want to waste time again with an answer that's not right for you.

[sandhya sandhya]

Thank you for your reply.your answer suitable to my question.No wastage of time happened to you for my question.But you asked about how i removed the trends,because of that i posted the link(remove trends from data).Try this link REMOVE TRENDS FROM DATA or just type the title in matlab documentation.

[sandhya sandhya]

I posted another question,how to plot multiple mat signals in a single plot.I attached my mat signals to that question.please help me to plot.

[Image Analyst]

OK, maybe if the questions are numbered you will notice them more than if they just ended with question marks:

1. I also don't know what you define as a negative peak. Is it everything between positive peaks? If so, setting negative peaks to zero would just zero out the entire signal except for the very tips of the positive peaks, and that doesn't seem any good.
2. Is it the part of your signal that is negative (has a value below zero)?
3. When you detrend it, what functions and parameters to you use?
4. Did you detrend()?
5. Or did you use polyfit(), and if so, with what order?

[sandhya sandhya]

1.I want to set all those negative peaks to zero and its an optional.I want to work only on positive peaks.After that i want to calculate average amplitude of the signal.

2.I applied detrending technique based on the command detrend().If polyfit() command suits to my signal then apply that command to my signal.