The surface / surf plot in MATLAB can visualize data in 3D. When I took a course in grad school on statistical image processing, I saw a very interesting plot where data is shown as a surf
plot and underneath, on the ground or x-y plane, an image is shown. The pixels of the image corresponded to the points in the 3D surface and gave some extra information about the each point, sort of like an image-based version of surfc
or a 3D version of plotting over an image background. I always wanted to know how to make that plot but rather than asking the prof who made it (as one is supposed to after paying tuition), I decided to figure it out on my own (thus proving why I was never good at accounting). It took some experimentation but I finally figured out how this type of plot is accomplished. In this tutorial, I will show how to do this and how to make it so that the surface plot and the image can use different colormaps, getting around the restriction that MATLAB only has one colormap per figure. In effect, this will simulate multiple colormaps. The result will look something like this:
Note how the image is mapped to a plane and is shown with a jet
colormap while the 3d surf is shown with a gray
colormap.
I will first go through the basic way to do this without multiple colormap support. Take a look at this code:
% the data that you want to plot as a 3D surface. [x,y,z] = peaks; % get the corners of the domain in which the data occurs. min_x = min(min(x)); min_y = min(min(y)); max_x = max(max(x)); max_y = max(max(y)); % the image data you want to show as a plane. planeimg = abs(z); % set hold on so we can show multiple plots / surfs in the figure. figure; hold on; % do a normal surface plot. surf(x,y,z); % set a colormap (but this has no effect because the next colormap % command overwrites it) colormap(gray); % desired z position of the image plane. imgzposition = -10; % plot the image plane using surf. surf([min_x max_x],[min_y max_y],repmat(imgzposition, [2 2]),... planeimg,'facecolor','texture') % set a colormap for the figure. colormap(jet); % set the view angle. view(45,30); % labels xlabel('x'); ylabel('y'); zlabel('z');
The above code will produce this plot:
I first created the 3D data to be plotted by making use of the peaks
function. Next, I found the domain of the data (extent in the x-y plane). This is used to define the plane that the image will be mapped to in the final plot. The image to be shown is stored in planeimg
(in this case I simply used abs(z)
but the image can any arbitrarily sized array of data - it will be stretched or squeezed to fit the plane defined by the extent of the domain of the data). I then plotted the 3D data by calling surf
. Finally, the image is mapped to a plane parallel to the x-y plane by the following lines:
% desired z position of the image plane. imgzposition = -10; % plot the image plane using surf. surf([min_x max_x],[min_y max_y],repmat(imgzposition, [2 2]),... planeimg,'facecolor','texture')
imgzposition
sets where on the z axis the image plane is to be placed; I set -10 so it would not intersect with the existing 3D surface. The surf
command here has defined a planar surface with the following vertices:
(min_x, min_y, imgzposition)
(max_x, min_y, imgzposition)
(max_x, max_y, imgzposition)
(min_x, max_y, imgzposition)
To understand how this works, refer to MATLAB's surf documentation, which states:
surf(X,Y,Z) creates a shaded surface using Z for the color data as well as surface height. X and Y are vectors or matrices defining the x and y components of a surface. If X and Y are vectors, length(X) = n and length(Y) = m, where [m,n] = size(Z). In this case, the vertices of the surface faces are (X(j), Y(i), Z(i,j)) triples.
In any case, the planeimg
is used as a texture map for the single face of the planar surface.
You will have noticed that I actually have two colormap
commands in the code, but both the 3D surface and the image have the same colormap. This is because there is only one colormap for the figure. The next section will discuss how to get around this and use a different colormap for the 3D surface and the image plane.
To give the image a different colormap than the 3D surface, all I need to do is convert the image (which is just a 2D array of values) into a true color image according to the desired colormap. This is how:
% scale the between [0, 255] in order to use a custom color map for it. minplaneimg = min(min(planeimg)); % find minimum first. scaledimg = (floor(((planeimg - minplaneimg) ./ ... (max(max(planeimg)) - minplaneimg)) * 255)); % perform scaling % convert the image to a true color image with the jet colormap. colorimg = ind2rgb(scaledimg,jet(256));
Due to the way colormap
and ind2rgb
work, the image must first be scaled to between [0, 255]
before being converted to true color. I also have to specify a 256 element colormap (since [0, 255]
has 256 possible values).
Now all we have to do is instead of showing planeimg
, we show colorimg
instead:
% plot the image plane using surf. surf([min_x max_x],[min_y max_y],repmat(imgzposition, [2 2]),... colorimg,'facecolor','texture')
This will produce a figure very similar to the one at the beginning of this tutorial. If you want to get rid of the wireframe mesh (as I did), then simply specify 'edgecolor' to be 'none' in your surf
commands:
surf(x,y,z,'edgecolor','none');
Here it is altogether:
% the data that you want to plot as a 3D surface. [x,y,z] = peaks; % get the corners of the domain in which the data occurs. min_x = min(min(x)); min_y = min(min(y)); max_x = max(max(x)); max_y = max(max(y)); % the image data you want to show as a plane. planeimg = abs(z); % scale image between [0, 255] in order to use a custom color map for it. minplaneimg = min(min(planeimg)); % find the minimum scaledimg = (floor(((planeimg - minplaneimg) ./ ... (max(max(planeimg)) - minplaneimg)) * 255)); % perform scaling % convert the image to a true color image with the jet colormap. colorimg = ind2rgb(scaledimg,jet(256)); % set hold on so we can show multiple plots / surfs in the figure. figure; hold on; % do a normal surface plot. surf(x,y,z,'edgecolor','none'); % set a colormap for the surface colormap(gray); % desired z position of the image plane. imgzposition = -10; % plot the image plane using surf. surf([min_x max_x],[min_y max_y],repmat(imgzposition, [2 2]),... colorimg,'facecolor','texture') % set the view. view(45,30); % label the axes xlabel('x'); ylabel('y'); zlabel('z');
This will produce the figure shown at the beginning of the tutorial:
Discussion
I used it to combine eye-tracking data with the picture which was watched.
But I have problems exporting it from MatLab. "saveas" and "print" just export the surf but not the picture.
Did anybody try this?
% png
saveas(gcf, 'output.png');
% pdf
saveas(gcf, 'output.pdf');
An alternative to try if saveas is somehow messed up is to use getframe to capture the current figure as an image and then write it out using imwrite:
c = getframe(gcf);
imwrite(c.cdata, 'output.png');
Now it works. I used the same command before but it didn't work.
Maybe there was an error in my script...
this is very useful for us......i need one help.I need matlab coding for 3d image
How to mark different points in different surface plots;
I have just used surface plot and found all my labels (titles, legend, tick-marks...) upside down and mirrored (i.e. if you want to read them correctly you would have to stand on your head behind the screen). Has anyone ever seen anything like this? Why does it happen, and how do I fix it?
http://www.mathworks.com/matlabcentral/newsreader/view_thread/243468
I was wondering, how to manipulate the code, if one want to do a 3D surface plot for some image that one has?
Can you please help he out of this?
I want to give the flat surface plot a grid-like appeareance (such as the one from the 3D surface). I have tried with the commands MeshStyle (both), LineStyle (-,--), and FaceColor (flat, interp). None of these commands gave me the desired result. Actually, changing the value for FaceColor made the flat surface plot dissapear (Matlab gave the following comment: Warning: size(CData) must equal size(ZData) or size(ZData)-1 for flat shading).
Do you have any suggestions for this task?
Thanks in advance.
http://www.mathworks.com/help/techdoc/ref/surface.html#f67-521436
Instead of using the peaks function as in that example, you should be able to create a flat grid surface.
if i want to make this plot for a FMRI image then what should i do?
Do you have any suggestions for this task?
Thanks in advance.
please note that as an academic project we have decided to chenge a 2D image of face to a 3D image to machinhig by CNC through matlab software.kindly sbmit your comment or references that can help us in this project.
can you help to plot an mirror image of an object using MATLAB.
% the data that you want to plot as a 3D surface.
[x,y,z] = peaks;
How can I utilize your code for my 2D color image? is it by using imread function? Hope you could help me on this matter
Thanks.
Great code it was usefull for the first part of my project.
But I would like to know if it is possible to plot the 2D image in the background of a scattered volume (for which I use scatter3) and make the 2D image static, in order to be able to rotate the volume, without rotating the 2D image, to find the position of the volume that best matches the 2D image and then obtain a projection of the volume to overlay over the 2D image.
In this exact same setting, I am struggling to display two colormaps corresponding to the 3D plot and the surface plot. Do you know of a way? Thanks so much.