3D curve line fitting

Hi all,

My basic question is this. I have a set of data in 3D space that makes an approximate ellipse ring, with a few bends and noise. Is there an easy way to connect the points in a smooth way to make a nicely connected ring?

I created this simple code to represent what my data may look like.

%Create X and Y based on ellipse equation
x = -2:.1:2;
xnoise = (rand(1,81)-0.5)/10;%Generate small noise in x
y = sqrt(1-((x.*x)/4));
ynoise = (rand(1,81)-0.5)/10;%Generate small noise in y
x = [x(1:end-1) flipdim(x,2)];
x = x+xnoise;
y = [-y(1:end-1) y];
y = y+ynoise;
%Create Z using sine
s = 0:2*pi/80:2*pi;
z = sin(s);
znoise = (rand(1,81)-0.5)/10;%Generate small noise in z
z = z+znoise;
%Plot data field
plot3(x,y,z,'o')

Keep in mind that my knowledge of Matlab is limited and I would like a solution that can be utilized in coding as opposed a toolbox function. Thanks in advance!

Answer 1

[kedija]

I asked the same question and get a nice answer though it has past a year since you ask it might help others to post the answer here.

just append the following at the end

hold on; 
d= [x;y;z]'; 
[th, r, z] = cart2pol(d(:,1),d(:,2),d(:,3)); 
[sr, idx] = sort(th); 
sd = d(idx,:); 
plot3(sd(:,1), sd(:,2), sd(:,3))