how to calculate wavelength of texture ?

I have the texture the wavelength as in the figure attachment. I want to calculate the distance from one peak to one peak as show in the figure. How can I calculate it ? thank you very much

Answer 1

[Star Strider]

Consider using the smoothdata function to eliminate the high-frequency noise. After that, first, use the detrend function to remove the varying baseline (this makles the peak amplitudes and location determinations more reliable), then use findpeaks or islocalmax to locate the peaks. (There are other options if detrend does not produce acceptable results, however it is always the first option I use.)

[Lyhour Chhay]

Dear Sir,

Thank you very much for you reply. I used it based on your suggestion already. It does not produce acceptable results. I will show you in the figure. (blue is original, orange is smoothdata function).

[Star Strider]

The smoothdata function has several parameters (specifically window length) that can be varied to produce a result with significantly decreased noise whilst preserving the importaant features of the rest of the signal.

The findpeaks and islocalmax functions allow a specific prominence value (and another to define the minimum distance between detected peaks, if necessary) to be stated that will greatly increase the probability of detecting only the correct peaks and locations.

I do not have the data, so I can only describe the approach, and cannot provide specific code.

[Lyhour Chhay]

Dear Star Strider,

Thank you very much sir. I will send you the .mat file in the given link below. Actually, I am extract the surface profile from the point cloud. I use the code below to extract from the point cloud and calculate the peak distance. However, I notice that it is not so well for cross section profile extraction. It would be better if you could help me to extract the profile from the point cloud with the given coordinate y then I can plot the profile (x,z). Then, I can calculate the peak distance from point cloud more accurately. I am sorry for you inconvienience sir. Thank you very much sir.

https://gwnuackr-my.sharepoint.com/:u:/g/personal/lyhour_gwnu_ac_kr/EWXF-iarV9lLp2GIBiAbjyEBNPnuOAHANu886wzMt8eEwQ?e=3mDJir

clc,clear
ply=pcread('result.ply');
x=ply.Location(:,1)*1000;
y=ply.Location(:,2)*1000;
z=ply.Location(:,3)*1000;
xmin = ply.XLimits(1)*1000;
xmax = ply.XLimits(2)*1000;
ymin = ply.YLimits(1)*1000;
ymax = ply.YLimits(2)*1000;
%% Extract surface profile
F = scatteredInterpolant(x,y,z,'linear','nearest');
xp = linspace(xmin,xmax,length(F.Points));
yp = linspace(ymin,ymax,length(F.Points));
xp = xp';
yp = yp';
zp =F(xp,yp);
%% MWL calucation
[pks,locs] = findpeaks(zp,xp,'MinPeakDistance',3);
plot(xp,zp)
hold all
plot(locs,pks,'ro','MarkerSize',10)
MWL = mean(diff(locs));
fprintf('\n Mean wavelenght is %f \n', MWL); 

[Image Analyst]

We're not going there. You can attach the mat file here with the paperclip icon. If it's bigger than 5 MB, then crop down your data to a smaller size.

[Lyhour Chhay]

Dear Sir,

Thank you very much sir. Glad to see you again. Does my link provide not working ? My mat file is bigger than 5MB. How can I crop down my data ? you mean I have to reduce my point cloud ?

[Image Analyst]

Yes. Like if it's an array called data that is 10 MB, cut it in half. Let's say that's only the first 3000 rows, then do

dataCropped = data(1:3000, :);
save('answers.mat', 'dataCropped');

[Lyhour Chhay]

Dear Sir,

I crop the point cloud already the mat file that I save is only the point cloud. It is very short length I am concern that it is effect to the calculation of wavelenght and data extraction. I attach it already.

[Star Strider]

When I attemtp to load that, I get:

LD = load('dataCrop.mat');

plyRoi = LD.plyRoi

Warning: Variable 'plyRoi' originally saved as a pointCloud cannot be instantiated as an

object and will be read in as a uint32.

plyRoi =

6×1 uint32 column vector

3707764736

2

1

1

1

1

This does not appear to be the data in the plot.

Please attach the correct array. It might be best to try it as a text file instead.

.

[Lyhour Chhay]

Dear sir, The original version is store in this link: Data

However, if you want to use the Datacrop.mat file: - the data x,y,z coordinate are extracted from this command: x=plyRoi.Location(:,1)*1000; y=plyRoi.Location(:,2)*1000; z=plyRoi.Location(:,3)*1000; - and I use the code below to do the surface profile. I explained in the previous comment already. I will show you again sir.

x=plyRoi.Location(:,1)*1000; 
y=plyRoi.Location(:,2)*1000; 
z=plyRoi.Location(:,3)*1000;


% Extract surface profile 
F = scatteredInterpolant(x,y,z,'linear','nearest'); 
xp = linspace(xmin,xmax,length(F.Points)); 
yp = linspace(ymin,ymax,length(F.Points)); 
xp = xp'; 
yp = yp'; 
zp =F(xp,yp); 
%% MWL calucation 
[pks,locs] = findpeaks(zp,xp,'MinPeakDistance',3); plot(xp,zp) hold all 
plot(locs,pks,'ro','MarkerSize',10) 
MWL = mean(diff(locs)); fprintf('\n Mean wavelenght is %f \n', MWL);

[Star Strider]

It took me a while to sort all this out.

I changed the findpeaks call to:

[pks,locs] = findpeaks(zp,xp,'MinPeakProminence',0.4, 'MinPeakDistance',3.5);

that appears to be the only change required, and that then resulted in:

and:

Mean wavelenght is 7.380249

No further smoothing or other processing was necessary.

That may need to be tweaked for other data sets. It appears to work here.

.

[Lyhour Chhay]

Dear Sir,

Thank you for your much effort. I really appreciate it. please apology me for the late reply due to different time zone. I understand your suggestion and methodology. Due to my extraction profile is not well smooth, thus, my calculate of mean wavelenght also not so accurate. Therefore, I have to find the good method for profile extraction from point cloud. However, the method that you proposed for calculation between peak distance is acceptable. Thank you very much sir.

[Star Strider]

As always, my pleasure!