How to extract the X and Y values from a contourf graph

I am trying to generate a list of X and Y coordinates following the outline of a contourf graph. However, when I attempted the solutions that I found in other questions it resulted in an empty variable. Attached is my code to generate the contour graph and the resulting figure, thank you in advance for your help.

FinalMatrix = [AA AB AC AD AE AF AG AH AI AJ AK AL];

asdf = max(FinalMatrix,[],'all');

asdf = asdf*.25;

FinalMatrix(FinalMatrix<asdf) = NaN;

x = 1:12;

y = 1:7;

[X,Y] = meshgrid(x,y);

contourf(X,Y,FinalMatrix,[1 1])

[F,L] = contourf(X,Y,FinalMatrix,[1 1]);

[the cyclist]

It is unclear to me what you have as input. Do you have only the image (e.g. a *.jpg file?), or do you have the MATLAB *.fig file? (I assume that you do not have the original data used to create the figure, or it would be trivial.)

Can you upload what you have? You can use the paper clip icon in the INSERT section of the toolbar.

[Aidan Fishenden]

I do have the data used to create the figure, I've attached it here. Sorry, this is my first project with matlab.

Answer 1

[the cyclist]

The problem is that you set the levels input to contourf to [1 1], but it is impossible to have contours there, because all your data are greater than 1. So, it makes sense that there are no contours.

If you don't set that (or set it to values where contour heights are possible), you will see the contour output.

load("matrix.mat","FinalMatrix")

x = 1:12;
y = 1:7;
[X,Y] = meshgrid(x,y);

figure
[F,L] = contourf(X,Y,FinalMatrix);

F

F = 2×129

40.0000 2.0161 2.0127 2.0000 1.7500 50.0000 5.7097 6.0000 50.0000 2.1774 2.1392 2.0000 1.4118 50.0000 5.0000 4.9012 60.0000 5.3871 6.0000 60.0000 2.3387 2.2658 2.0000 2.0000 60.0000 6.0000 5.8182 5.8947 60.0000 5.0000 4.0000 2.0161 2.0000 1.9524 1.7500 2.0000 1.0000 1.2500 4.0000 2.1774 2.0000 1.4762 1.0000 2.0000 4.8462 5.0000 2.0000 1.0000 1.5278 4.0000 2.3387 2.0000 1.0000 1.0000 3.0000 2.8095 3.0000 3.1053 2.0000 4.6538

[Aidan Fishenden]

Thank you. That part works, but I am also trying to get only the coordinates of the outermost edge, not every value on the contour.

[the cyclist]

OK, but that really has nothing to do with the contour command, and really only has to do with the pattern of NaNs that you put in place.

I think the definition of the outermost edge is a bit tricky. Here are some thoughts.

It is like you have several "islands" of non-NaN values, and you want the "coastline" (what you call the "outermost edge") of those islands. I would say that you therefore want to exclude the following:

• All NaN values
• Non-NaN values that are fully surrounded by NaN values

This leaves some tricky cases. For example, the value FinalMatrix=40 is sitting by itself, surrounded by NaN. Does that count as its own "outermost edge"? I would think yes, but it might not, depending on what you need. There are similar issues for other smaller "islands".

It feels like maybe we are falling into the XY Problem, where we are fixing your solution, when there may be a better approach to solving your actual problem.

[Aidan Fishenden]

Possibly, but to answer your question no, I am not counting any "islands". I am only interested in the coastline of the main spike. This data is generated from a neuron, and the random pieces of data are from additional neuronal noise. If the contour function will not work for this, do you have any sugestions for something else that may help? To further clarify, I do not need the value that is in the cell, only the x and y value. For example, for this graph the values I would need are (1,1) (2,2) (3,3) (3,4) (3,5) (3,6) (4,5) (5,5) (5,4) (6,3) (6,2) (6,1)

[the cyclist]

OK. One thing that still makes this difficult is that we don't know enough about how general this needs to be. For example, is the spike always at the upper edge of the matrix, or might it have edges all around? We don't know what assumptions are valid.

There is probably not a simple MATLAB function to do what you want. Maybe something in the Image Processing Toolbox would make this easy, but I am not personally knowledgeable about it.

What I suggest is, before you think about coding this, is to think about the algorithm (the "instructions") that the code needs to follow. Think about each of the steps. For example, the first step might be to identify the spike itself, rather than any other values. How would you tell another person how to do that? That will help you figure out the rule, which can then be coded.

We could write code based on assumptions and guesses, but it will probably be wrong.

[Aidan Fishenden]

Understood, thank you.