Filled contour/colorized 2-D surface plot with large 3-D array (10000 x 3)

Hi!

Can you please help me generate a filled contour plot and/or a colorized 2-D surface plot using a large 3-D array (10000 x 3; see attached file)? The data is arranged in the following format: x-coordinate, y-coordinate and density.

I am hoping to visually end up with something like this (i.e., with minimal interpolation):

Example image

Thanks so much!

Answer 1

[Aditya]

Thanks for your help Mathieu! I managed to solve this problem using imshow by looking at previous Matlab answers by Image Analyst:

x = D(:,1);

Unrecognized function or variable 'D'.

y = D(:,2);
z = D(:,3);
xgrid = reshape(x,[100,100]);
ygrid = reshape(y,[100,100]);
zgrid = reshape(z,[100,100]);
C= [xgrid ygrid zgrid];
C2 = imresize(zgrid, 100); % Blow up by a factor of 100;
figure;
imshow(C2,'XData', x./1e3, 'YData', y./1e3);
axis on;
clim([0e14 0.16e14])
cmap = jet(256);
colormap(cmap);
h = colorbar;
h.Label.String = 'Number Density';
h.Location = 'northoutside';
xlabel('Km'); ylabel('Km');
set(gca,'Fontsize',16);
ax = gca;               % get the current axis
ax.Clipping = 'off';    % turn clipping off

Answer 2

[Mathieu NOE]

hello

maybe this ?

your data is large but lacks some refinement in the central area of the peak

90% of your data is Z close to zero

3D rendering

if we make a flat 2D rendering you will see that we have a very coarse grid in the center area

zoomed here :

load('test_data.mat')
x = D(:,1);
y = D(:,2);
z = D(:,3);

%% Plot it with TRISURF
tri = delaunay(x,y);
h = trisurf(tri, x, y, z);
set(h,'LineWidth',0.01)
shading interp
% view(2); % uncomment to have a flat 2D rendering
colorbar EastOutside

[Aditya]

Thank you so much! However, the grid is actually uniform (not coarse in the center), as can be seen by plotting scatter(x,y). I've tried using your solution, pcolor and imagesc, but all of them seem to show 'wedge'-like shapes near the origin.

Do you know how to ensure that the interpolation is done smoothly?

[Mathieu NOE]

sorry if I was unclear

I didn't mean that the resolution was coarser in the center, I just mean that the overall resolution is too coarse to give a smooth rendering of what is supposed to be displayed

also , we could reduce the number of points as outside the very center portion, the rest is almost zero in z direction, so not very interesting

yes , if you display the data with scatter, you can see that the number of points where the z data is non zero is very limited

you may try to improve with interp2 but don't do it on the full data set , simply on the center area

how were the data actually generated ?