How to extract data from figure in matlab?

Question

I have saved different Matlab plots in an unique .fig. The figure is like this: Now, I would like to introduce a filter in these plots to reduce the noises, but unfortunately I have lost the code that generates these signals.
Is there a way to extract data of each signal in this figure? I tried this:

open('ttc_delay1000.fig'); 
h = gcf; %current figure handle
axesObjs = get(h, 'Children');  %axes handles
dataObjs = get(axesObjs, 'Children'); %handles to low-level graphics objects in axes

objTypes = get(dataObjs, 'Type');  %type of low-level graphics object

xdata = get(dataObjs, 'XData');  %data from low-level grahics objects
ydata = get(dataObjs, 'YData');

But I am confused and I don't know if it's the right way to act. Thanks!

Kinsfolk

Well, what does your code produce? – Cambium

I have Objtypes and xdata, ydata, structures composed by a 5x1 cell. Each row of these structures contains the data of each signal, i guess. Is that right? – Kinsfolk

It would seem so:) Try reproducing your plot on a new figure, based on those data. That's a sure way of knowing for sure. Note that handles are usually stored in vectors in a reverse order: stuff plotted last usually pop to the front of the list. – Cambium

yes, this code should do the job; but remember that you've lost your function and only have finite number of samples of your data, so you want to be smart about how you want to filter/reduce noise. – Ectomere

indeed it works writing plot(xdata{1,1}, ydata{1,1}) and so on. Yes, now I have to understand how to filter these signal. But I think it 's possible.. – Kinsfolk

I guess the answer that I posted is not what you wanted... could you be more specific on what is the expected output? And update the question accordingly, while we're at it? – Kalidasa

Actually yes, my code works and also yours, that is way more compact. Maybe it's better to update the question :) – Kinsfolk

A brutal way to filter your high frequency noise (I call it the chuck Norris) is : 1/ run an fft on your signal - 2/ set to 0 all coefficients corresponding to high frequencies 3/ run ifft on the result to get filtered signal. Mathematicians or non Chuck Norris fans would strongly recommend using a Gaussian low pass filter after the fft step – Rooster

@Kinsfolk I've added code to smooth out the lineseries – Kalidasa

Your original question about getting the data from the figures has been answered; asking about the filtering is an entirely different issue. Please open another question for that. – Biliary

Answer 1

A one-liner for your problem:

data = get(findobj(open('ttc_delay1000.fig'), 'Type','line'), {'XData','YData'});

The steps are there (from the inner calls to the outer calls):

• open the file;
• look into it for the line series;
• return the data.

data{n,1} will contain the XData of the LineSeries number n, wile the data{n,2} will contain the YData of the said LineSeries.

If you want to smooth the lines directly in the figure, the idea is the same:

    %//Prepare moving average filter of size N
    N = 5;
    f = @(x) filter(ones(1,N)/N, 1, x);

    %//Smooth out the Y data of the LineSeries
    hf = open('ttc_delay1000.fig');
    for hl = transpose(findobj(hf,'Type','line'))
            set(hl, 'YData', f(get(hl,'YData')));
    end;
    saveas(hf, 'ttc_delay1000_smooth.fig');

Kalidasa

Wait, findobj works for grandchildren? What have I been doing all these years! – Rooster

@Rooster When the first argument is a graphic handle (or an array of), the search is restricted to those handles and their descendants indeed. – Kalidasa