How to do subplot using cwt()

Hey, I would like to plot a 3*3 subplots containing 9 figures from cwt(). However, I found it impossible to do that by simply doing

subplot(3,3,1)
cwt(data1)
subplot(3,3,2)
cwt(data2)

My guess is that we have to manually extract the cwt features and plot by ourselves under the subplot(). If yes, could anyone share a sample code for this part? If not, what should I do to draw such a subplot?

Thank you very much!

[Walter Roberson]

Seems to work for me? Note that inputs must be vectors.

data1 = randi([0 255], 1, 128);
data2 = sort(data1);
subplot(3,3,1)
cwt(data1)

subplot(3,3,2)
cwt(data2)

[Shengjie Gao]

Hey Walter,

Thank you for your quick reply. The code works but it did not show a 2 by 1 layout with each scalogram created by using cwt(). It only shows the last one, here in your case it will only deliver data2's result. I would like to show both data1 and data2's cwt result in one figure (2-by-1 layout) here. Could you tell me how to do that?

[Walter Roberson]

Ah, I see what you mean. The code specifically clears the current figure before plotting.

In some cases, the code creates two axes, but I have not figured out yet which cases that corresponds to.

The code is taking the wavelet transform information (what would normally be the first output) and abs() it, and effectively does an imagesc() of that.

Answer 1

[Monisha Nalluru]

Hi Shengjie,

cwt() would support subplots only with complex valued input signals that will plot the analytic and anti-analytic parts into seperate subplot.

In all other case, the output of cwt can be used with surface,pcolor or image to produce similar plot

As an example

load mtlb
figure
subplot(211)
[wt1,f1] = cwt(mtlb,'bump',Fs);
pcolor(1:numel(mtlb),log2(f1),abs(wt1));
shading interp; 
subplot(212)
[wt2,f2] = cwt(mtlb,Fs);
pcolor(1:numel(mtlb),log2(f2),abs(wt2));
shading interp

Example cwt supporting subplots for complex valued iput ,

load npg2006;
plot(npg2006.cx); hold on; grid on;
xlabel('Eastward Displacement (km)');
ylabel('Northward Displacement (km)');
plot(npg2006.cx(1),'^','markersize',11,'color','r',...
'markerfacecolor',[1 0 0 ]);
figure;
cwt(npg2006.cx,npg2006.dt);

Hope this helps!

[Shengjie Gao]

Great! That works! Thank you very much!

[Shengjie Gao]

Hey Monisha,

Just a quick follow-up. Could you tell me how to modify both the xy-axis to the time-frequency range and also add the colorbar based on the magnitude?

Really appreciate your help!

Answer 2

[Jeremy Scholze]

Option 1: My Solution

Use

[cfs, frq] = cwt(data,Fs);
tms = (0:numel(data)-1)/Fs;

to calculate the wavelet transform and then do

imagesc(tms,frq,abs(C)); c = colorbar; c.Label.String = 'Magnitude';
axis tight; shading flat;
xlabel('Time (s)')
ylabel('Frequency (Hz)')
set(gca,'yscale','log')

to generate the plot. Your axes will be screwed up and you won't have the frequency bounds area greyed out but it'll work.

Option 2: MATLAB's Solution (less efficient)

openExample('wavelet/PlotCWTScalogramInSubplotExample')

They address this issue. Just run this in your command window ^

Edit: My method is apparently WAY more efficient than using the surface command in MATLAB's solution. Would recommend imagesc before surface. surface crashed my computer. The axis controls in MATLAB's solution will still work with imagesc.