advanced Heatmap plotting question

hi matlab experts, i have the following data matrix (see attached data.mat file)

i don't care about the all 0 column.

first column is x coordinate, second is y coordinate.

4th column is the value

as you can tell this is some sort of a scan that covers a 30x30 xy area

so for example, the first row of this data is -15,-15,0, 2921. it means, at x= -15 and y= -15, the data value is 2921 (ignore the 0)

question:

1) how do you plot a simple heatmap using this data?

2) more difficult: i really care about the values that are less than 3000. for me, any value less than 3000 means 'valid', any value above 3000 is invalid. so in addition to a normal heatmap that will just display the values in a temperature sense (like from cool to hot) , i really want another graph that clearly shows me what areas are 'valid' and what areas are 'invalid' in a black and white color sense (or green/red, whatever, as long as its just two colors) . one that i could easily look at and say, hey, at x=5 y=5, the value is invalid (above 3000)

thanks a lot!

Answer 1

[Alan Stevens]

The following will do it, though there are probably neater ways to do the colouring:

load('data.mat')
x = Z0(:,1); y = Z0(:,2); z = Z0(:,4);
x = reshape(x,31,31); y = reshape(y,31,31);z=reshape(z,31,31);
zlo = z<=3000;
C = 100*zlo;
surf(x,y,z,C)

[Alan Stevens]

By looking from the top you get:

[chang liu]

wow this is neat Alan.

is there a command that makes the figure just show up in a 2D fashion? (right now have to use mouse and drag it around a bit to rotate to the right view)

thanks

[chang liu]

hmmm... i found that by doing a

figure;hold on;

before you do the surf, it makes it into 2D

[Alan Stevens]

Great!

[chang liu]

hey... this may be asking too much but.. do you know a way to overlay a circle onto the existing heatmap? like i've manually done here. just a circle having certain radius and center, i don't need the 'x' markers or any other annotations that i've drawn here. they just there to help me draw this circle manually.

the reason is i need to very quickly gauge using my eye (out of hundreds of figures) if the valid area (yellow zone) meets certain minimal requirement such as having a minimal radius (bigger is ok) . sort of like a 'limit line' in converntional xy plots.

attached file as example

also, figure is already 2D, as i did the 'figure;hold on;' thing to create the figure in the first place.

[Alan Stevens]

Try this:

load('data.mat')
x = Z0(:,1); y = Z0(:,2); z = Z0(:,4); 
x = reshape(x,31,31); y = reshape(y,31,31);z=reshape(z,31,31);
zlo = z<=3000;
C = 100*zlo;
surf(x,y,z,C)
xlabel('x'), ylabel('y')
view(180,-90)
hold on
theta = zeros(360,1);
for i = 1:361; theta(i) = (i-1)*pi/180; end
r = 10;  % You will have to decide this yourself
xc = r*cos(theta); yc = r*sin(theta);
plot(xc,yc)

[chang liu]

thanks! worked like a charm

[chang liu]

alan

found another issue. so whenever the data is entirely below the threshold we set (3000), the color goes to cyan. can you help? attached sample data

ideally it should be all yellow.

[Alan Stevens]

If you add the colormap command just before the surf command like so:

...
C = 100*zlo;
colormap([0 0 0; 1 1 1]);
surf(x,y,z,C)
...

you will just get two colours (or just one for some situations!) .

The colormap above will result in black (the [0 0 0]) and white (the [1 1 1])..

By changing the values in the second row (i.e. the [1 1 1] values) you can adjust the colour (each of the digits must be between 0 and 1: they are RGB values). For example [127/255 1 212/255] is aquamarine (I can't think what yellow is off the top of my head).

[chang liu]

worked like a charm