How to find number of times graph goes back to zero

Hello everyone, I am new to MATLAB and I am currently plotting a normalised gyroscope data as in the picture below, and I am trying to count the number of the graph trying to go back to zero (on the word, number of the graph descending after reaching each peak)

For example, if I calculate it manually fromthe graph, it's around 9 times it tries to go back to zero. But I am trying to do analysis on big data so I really ned way of finding it by MATLAB.

Help me, thanks

Answer 1

[Star Strider]

To find the troughs, invert the ‘y’ values and use findpeaks:

Example:

x = linspace(0,250,250);                    % Create Data
y = 0.25*sin(0.25*pi*x) .* exp(-(x-125).^2./1000) + exp(-(x-125).^2./1000);

[pks,plocs] = findpeaks(y);                 % Find peaks
[trs,tlocs] = findpeaks(0.1+max(y) - y);    % Find troughs

figure(1)
plot(x, y)
hold on
plot(x(plocs), y(plocs), '^r', 'MarkerFaceColor','r')              % Plot peaks
plot(x(tlocs), y(tlocs), 'vg', 'MarkerFaceColor','g')              % Plot troughs
hold off
grid

produces:

[Sharah]

thanks for your suggestion. but the when it find the through, it still detect the line at the begining of the graph, and I don't need that. What is your fast suggestion for this?

[Star Strider]

The findpeaks function has ‘name’-‘value’ pair options. Here, 'threshold' is your friend.

With your data (assuming your y variable is ‘y’:

trhd = 1;                                                     % Set threshold
[pks,plocs] = findpeaks(y, 'threshold', trhd);                % Find peaks
[trs,tlocs] = findpeaks(0.1+max(y) - y, 'threshold', trhd);   % Find troughs

I don’t have your data, so just looking at your plot, I would start with a threshold (my ‘trhd’ variable) of 1 and increase or decrease it until you get the result you want.

[Sharah]

thanks. anyway, your attached picture looks so high in resolution and mine looks pixelated. how did you do that?

[Star Strider]

My pleasure!

Did the thresholding work as you want it to? If not, findpeaks has other options we can consider. If you want to post your data here I can experiment with findpeaks on it.

I always use the ‘.png’ format (portable network graphics). You used .jpg, which at least used to use a discrete cosine transform that produces ‘ringing’ at abrupt transitions. (If Image Analyst happens upon this, I will welcome his explanation and clarification.)

[Sharah]

thank you, that solution is very helpful. yes, i do have another problem with it but i posted on anther question http://www.mathworks.co.uk/matlabcentral/answers/134657-how-to-detect-point-before-it-went-on-plateau

[Star Strider]

My pleasure!

I saw your other post and suggested a solution.