Specify annotation position with respect to x- and y-axes values?

545 views (last 30 days)
Dear all,
Is there a way the define the starting and end point of an annotation object such as an arrow in terms of the data being plotted and not as either "normalized" coordinates (x- and y-positions between 0 and 1) or a physical length measurement such as cm?
E.g. I am plotting a time from 0 to 180 s on the x-axis and a concentration between 0 and 0.1 concentration units on the y-axis. Now I would like to draw an arrow starting at 90 s, 0.04 concentration units and going to 120 s, 0.09 concentration units.
I have already considered to simply "normalize" those positions myself (e.g. 90 s/180 s = 0.5 --> x-position in normalized coordinates). This does, however, not work, since 0 and 1 refer to points in the whole figure (the window in which the plot appears) and not to the area of the plot itself.
Thanks for any input! Clemens
  1 Comment
Clemens on 5 Nov 2016
Here's a quick illustration of what I mean. I simply drew the arrow by hand. If I get Matlab to show me the code for it, it gives
annotation(figure1,'arrow',[0.354235423542354 0.224422442244224],...
[0.526863777089783 0.390092879256966]);
but the end point's x-coordinate (0.2244...) does not correspond to the 20 s (20 s / 180 s = 0.1111...) where 180 s is the "length" of my x-axis in the units that I want to plot. 

Sign in to comment.

Accepted Answer

Star Strider
Star Strider on 5 Nov 2016
You need to access the 'Position' property of the figure object you’re using. It gives the left lower corner coordinates and the width and height values. With a few lines of code, you can normalise your arrow object positions with respect to them. See the documentation on Axes Properties for details.
Then, experiment! This will keep you intensely occupied for a few minutes!
Star Strider
Star Strider on 20 Sep 2017
More information would be helpful, as would more illustrative example code, since we don’t know what you’re doing, and providing a context is always appropriate. I assume ‘x_data’ is your independent variable vector. Also, is this robust to ‘x_data’ having negative values?
‘DrBones’ — Ortho? B/C IM here!

Sign in to comment.

More Answers (2)

Jorg Woehl
Jorg Woehl on 1 Mar 2021
Edited: Jorg Woehl on 1 Mar 2021
This is an older post, but the following is perhaps useful to others:
If horizontal arrows are ok, data units can be used by associating the annotation with the current axes:
hf = figure;
x = 0:180; f = @(x) 0.09*exp(-x/18);
plot(x, f(x));
ha = annotation('arrow');
ha.Parent = hf.CurrentAxes; % associate annotation with current axes
% now you can use data units
ha.X = [50 20];
ha.Y = [f(20) f(20)];
This also works for other annotation types and even if they don't have a horizontal orientation. (Arrows and doublearrows are an exception because the arrowheads will not align properly.)
Otherwise, there is also the excellent coord2norm function on File Exchange that converts data units to normalized units.

Sign in to comment.

DGM on 15 Aug 2021
As a spin off @Jorg Woehl's suggestion, I'll offer my own.
I think it's worth pointing out that these solutions tend to be version-dependent -- at least dependent on whether we're post-R2014b or not. Obviously, we can't manipulate properties of annotation objects using dot notation in legacy versions, but even using get/set to move the annotation tends to result in it being drawn in unexpected ways.
But precalculating the position and specifying it when creating the annotation works fine from R2009b to R2019b:
x = 0:180; f = @(x) 0.09*exp(-x/18);
plot(x, f(x));
[xx yy] = datc2figc([50 20],[f(20) f(20)]);
I used the attached file for coordinate conversion, but I doubt it's much different than coord2norm().
DGM on 15 Aug 2021
datc2figc() was really just an ad-hoc thing I threw together in frustration. I'm sure there's better stuff out there.

Sign in to comment.

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by