Locating points on the figure that correspond to a specific number on the colormap

I've opened a .m figure that was generated earlier.

openfig('MapSurface.fig')

Next, I would like to know the xy position of the points on the figure that exactly match with 100 on the colorbar.

This could either be a pixel or list of pixels (as per their xy position) that correspond to 100, or, a marked region on the figure that highlights all points corresponding to 100.

Note: I'm doing this as my final aim is to calculate the area of the spot size by knowing the location of its boundaries. In other cases the spot size is bigger or an arbitary shape. I've taken the simplest case for this question.

Thank you.

[Adam Danz]

Do you know how the image was created? imagesc()?

[Jigsaw]

I used the following code for it

A = [200 200 200 200 200; 200 200 200 200 200; 200 200 100 200 200; 200 200 200 200 200; 200 200 200 200 200];
Percent_matrix = flipud (A);

Percent_Interp = imresize(Percent_matrix,200,'bilinear');
pcolor(Percent_Interp)
shading interp
caxis([0 200])

(The colorbar is a custom one as the predefined ones in matlab didnt have the one i wanted)

Answer 1

[Adam Danz]

In this functional example below, the colorbar value (cval) is set to 140.0 which is within the yellow-green spectrum of the colorbar. The tolerance (tol) is set to 0.5 meaning that any value within 140 +/- 0.5 is accepted. A tolerance of 0 only accepts exact matches which will likely fail.

The "CData" (color data) is pulled from the figure and the code identifies which coordinates are within tolerance to your selected color value. Black markers are plotted showing the matches.

figure()
A = [200 200 200 200 200; 200 200 200 200 200; 200 200 100 200 200; 200 200 200 200 200; 200 200 200 200 200];
Percent_matrix = flipud (A);
Percent_Interp = imresize(Percent_matrix,200,'bilinear');
pcolor(Percent_Interp)
shading interp
caxis([0 200])
colorbar(); 

cval = 140.0;     %colorbar value
tol = .5;         %tolerance (accepts all values tval+/-tol); use 0 for no tol.

% get color data from the only object on the axis
ax = gca(); %handle to axis that contains the data
cdata = ax.Children.CData; % assumes axis only has 1 object!
% Alternative: cdata = ax.Children.ZData; 

% determine which coordinates match selected color value
cidx = cdata >= (cval-tol) & cdata <= cval+tol; 

% get the (x,y,z) coordinates of the slected color regions
x = ax.Children.XData; % assumes axis only has 1 object!
y = ax.Children.YData; % assumes axis only has 1 object!
[yidx,xidx] = find(cidx);  % *
xSelect = x(xidx); 
ySelect = y(yidx); 

% mark selected units on figure
hold on
h = plot(xSelect, ySelect, 'ks', 'MarkerFaceColor', 'k','markersize', 2); 

% delete(h)

[Adam Danz]

Perhaps contourf() would be more useful to plot the data.

figure()
A = [200 200 200 200 200; 200 200 200 200 200; 200 200 100 200 200; 200 200 200 200 200; 200 200 200 200 200];
Percent_matrix = flipud (A);
Percent_Interp = imresize(Percent_matrix,200,'bilinear');
contourf(Percent_Interp)
colorbar(); 

If you use controurf(), you'll need to replace 1 line in my code.

cdata = ax.Children.CData;  %OLD 
cdata = ax.Children.ZData;  %NEW
%                   ^

[Jigsaw]

(Thank you so much for the detailed explanation and the %comments after every line. They help a lot)

The code works perfectly for the given matrix case. I do have an issue with other cases though.

I changed the A Matrix so that there are two '100' points. I assume the code would track all the '100' corresponding points on the figure and mark them.

A = [200 200 200 200 200; 200 200 200 200 200; 200 200 100 100 200; 200 200 200 200 200; 200 200 200 200 200];
cval = 100.0;

So, in this case it should be a horizontal line along bright green. For some reason the direction is different.

I tried with different A matrix values (like 100 at different or multiple points) and the orientation seems to be inconsistent for all. What might be causing this?

another example

A = [200 200 200 200 200; 200 200 200 200 200; 200 200 200 200 200; 200 200 200 200 200; 200 200 200 100 100];

[Adam Danz]

The problem was that the x and y indices were mixed up. I fixed 1 line of my code marked with

"% *".

[xidx,yidx] = find(cidx);     %old, incorrect
[yidx,xidx] = find(cidx);     %updated, correct

The reason this error went undetected in the first example was because the plot was symmetric.

[Jigsaw]

I tried your updated code for different cases and it works perfectly.

Thanks a ton!

Cheers.

[Jigsaw]

Oh and I like your suggestion of using contourf().

Its a more neater representation but instead of all the different colour patterns getting contours, if only points corresponding to 100 get a contour then that would do the job.

So basically it would be a representation of all the 100 corresponding points falling within the marked boundary.

[Adam Danz]

Check out the "levels" input to contourf() where you can set the number of contrours.