Plot a 3D matrix for a three-dimensional object

I have obtained a 3D matrix A(:,:,3) for the geometry of an object. A(:,:,1) is a matrix corresponding to x coordinates in the world coordinate system, and A(:,:,2) stores y coordinates and A(:,:,3) stores the z coordinates. So each point the object has three coordinates (x,y,z). It is noted that these coordinates are scattered, and some of them are non-integer.

Now I am going to plot the 3D surface or shape of the object. I am wondering what Matlab function or functions are best or suitable for the purpose.

A possible way is to use plot3 as follows:

 x = A(:,:,1);
 y = A(:,:,2);
 z = A(:,:,3);
 figure; plot3(x,y,z,'.-'); %the figure is attached below

But this function does not give a desired figure. less authentic, no shading or camlight functions supported. trisurf.m seems a good choice for the problem. Are there any other options?

Answer 1

[Chad Greene]

Hi Yuzhen,

Are the x and y values evenly spaced, or are they randomly scattered about? If the x and y values are evenly spaced, you can use xyz2grid.

[X,Y,Z] = xyz2grd(A(:,1),A(:,2),A(:,3)); 
surf(X,Y,Z) 
shading flat
camlight

But if the data points are scattered, you'll have to grid them with one of the scattered interpolation functions. My go-to gridding function is John D'Errico's gridfit, which is on File Exchange.

[Yuzhen Lu]

Thanks for your answer!

These coordinates are not evenly spaced, and some of them are non-integer. I will try your suggestion

[Chad Greene]

Are you sure there's no regularity to the x and y spacing? It's hard to tell in your plot, but it looks like there might be some common intervals. It's fine is some of them aren't integers, you can still use xyz2grid. I try to avoid any kind of grid fitting if the data are already gridded but inconveniently formatted.

Answer 2

[Arjun Tandon]

I don't have the xyz2grd function, so I've done it this way. It's not pretty, but it's worked for me.

%% Cleaning up 
clc
clear
close

%% Generating 3D data
for i = 1:100
    for j = 1:100
        for m = 1:100
            A(i,j,m) = i+j+m;
        end
    end
end

%% This is the important part. Here, I'm just extracting the 
%% vector data and storing it in variables.
X = A(:,1);
Y = A(:,2); 
Z = A(:,3);

T = [X,Y,Z];    %Creating a matrix (10 by 10 by 10) 

%% Graphical bits
surf(T)
shading flat

me. Hypothetically, you could also modify Chad's method (shown below); I was too lazy to try.

T = [A(:,1), A(:,2), A(:,3)];

Answer 3

[Swati Jain]

For future if someone come across same issue:

One can use pchow

https://www.mathworks.com/help/vision/ref/pcshow.html

and scatter3

https://www.mathworks.com/help/matlab/ref/scatter3.html

You can see shading with scatter3 by using the follwoing:

scatter3(x, y, z,[],z, '.')

Answer 4

[Nilar Htun]

x = A(:,:,1);

y = A(:,:,2);

z = A(:,:,3);

[x, y, z] = meshgrid(A(:,1), A(:,2), A(:,3));

surface(x,y,z);

figure;

scatter3(x,y,z,'.-');

figure;

shading flag