2D color map shows the image correct but the Y scale is flipped.

Hello Community,

I wrote a code for plotting 2D color map. Though it works and get the image of what I wnat but the Y scale is flipped. The Y axis is in logarithmic scale and shows the scale from positive to negative instead of negative to positive. I tried multiple ways to flip it but nothing happens. I have also attached an image of how the Y scale should look like.

% Prompt the user to select multiple data files 
[dataFiles, dataDirectory] = uigetfile('*.txt', 'Select data files', 'MultiSelect', 'on'); 

% Check if the user canceled the file selection 
if isequal(dataFiles, 0) 
    disp('No files selected. Exiting...'); 
    return; 
end

% Prompt the user to select an output directory 
outputDirectory = uigetdir('Select an output directory', 'Specify Output Directory'); 

% Check if the user canceled the output directory selection 
if outputDirectory == 0 
    disp('No output directory selected. Exiting...'); 
    return; 
end

% Initialize variables to store data
xData = []; 
yData = []; 

% Loop through selected data files 
for i = 1:length(dataFiles) 
    % Construct the full file path 
    fullFilePath = fullfile(dataDirectory, dataFiles{i}); 
    
    % Read the data from the current file 
    data = dlmread(fullFilePath); 
    
    % Extract x and y data (1st and 4th columns) 
    x = data(:, 1); 
    y = data(:, 4); 
    
    % Apply the formula to the y data 
    y = (10.^((y - 3.51307) / 0.22845)) ./ (7.75E-5 * 0.2); 
    
    % Append data to the storage variables 
    xData = [xData; x]; 
    yData = [yData; y]; 
end

% Define the number of bins for the histogram 
numXBins = 200; 

% Adjust as needed 
numYBins = 100; % Adjust as needed 

% Define the bin edges for the X and Y axes 
xBinEdges = linspace(min(xData), max(xData), numXBins); 
yBinEdges = logspace(log10(min(yData)), log10(max(yData)), numYBins); 

% Create a 2D histogram color map with specific bin edges 
h = histcounts2(xData, yData, xBinEdges, yBinEdges); 

% Create a color map using imagesc with the Y-axis and scale reversed 
figure; 
imagesc(xBinEdges, flip(log10(yBinEdges)), log(h)'); % Reverse the Y-axis and its scale 

colormap('jet'); % Use the 'jet' colormap for colorful visualization 
colorbar; % Display the color bar for the Z-axis (point density) 

% Customize the plot appearance 
xlabel('X-Axis (Column 1)'); 
ylabel('Log(Y-Axis)'); % Y-axis is logarithmic and reversed 
title('2D Histogram Color Map of Multiple Data Files'); 

% Specify the output image file path 
outputImageFilePath = fullfile(outputDirectory, 'histogram_colormap.png'); 

% Save the color map plot as an image (e.g., PNG) 
saveas(gcf, outputImageFilePath, 'png'); 

% Display the path where the image is saved 
disp(['2D Histogram Color Map saved to: ' outputImageFilePath]);

Editor's note: reformatted unreadable one-lined code block

Answer 1

[Dyuman Joshi]

Using the high level version of imagesc changes the direction of y-axis to 'reverse'.

Revert it back to normal -

C = [0 2 4 6; 8 10 12 14; 16 18 20 22];
imagesc(C)
set(gca, 'YDir', 'normal')

[Sneha Kandapal]

Thank you for replying. Do you mean I change this code "% Create a color map using imagesc with the Y-axis and scale reversed

figure;

imagesc(xBinEdges, flip(log10(yBinEdges)), log(h)'); % Reverse the Y-axis and its scale

colormap('jet'); % Use the 'jet' colormap for colorful visualization

colorbar; % Display the color bar for the Z-axis (point density)"?

[DGM]

When calling image()/imagesc()/imshow(), the XData, YData parameters are taken as 2-element tuples defining the range of x,y data respectively, and the image content will be distributed linearly between those extremes. If you pass vectors specifying logscale (or any arbitrarily-spaced) x,y positions, then everything except the extrema of those vectors will be ignored. The image data will not be spaced in log-scale.

So feeding a specific set of bin edges to the YData parameter will probably not do what you think it does, and flipping the sequence of bin edges won't change anything (since only the extrema matter).

Furthermore, your bin edges are already in log scale, but then you take the log of them again when you feed them to imagesc().

If you want to flip the direction of the x or y axis, do exactly as described above.

set(gca, 'YDir', 'normal')

If you find that your image is flipped with respect to the axis direction, you can use some combination of the above, along with flipud() on the image array (h).

If you want to define data locations in a nonlinear coordinate space, simply don't use image()/imagesc(). The easiest way to deal with that is to use pcolor() instead.

% an image with a nonlinearly distributed feature
n = 1000;
xrange = [0 2];
yrange = 10.^(xrange);
x = linspace(xrange(1),xrange(2),n);
y = linspace(yrange(1),yrange(2),n);
[X,Y] = meshgrid(x,y);
Z = exp(-(Y-10.^x).^2);

% as shown, it appears upside-down, since the origin is the NW corner
imagesc(X(:),Y(:),Z)

%% if you try to use imagesc() with nonlinear X,Y data
% only the extrema of X(:),Y(:) are used
% the data is distributed linearly between those extrema
% regardless of how you scale the axes
imagesc(X(:),Y(:),Z)
colormap(parula)
set(gca,'ydir','normal')
set(gca,'yscale','log')
xlabel('x'); ylabel('y (wrong)')

% the same thing, but trying to directly scale YData
imagesc(X(:),log10(Y(:)),Z)
colormap(parula)
set(gca,'ydir','normal')
set(gca,'yscale','linear')
xlabel('x'); ylabel('log_{10}(y) (wrong)')

%% just use pcolor()
% now the image content is distributed as dictated by the YData
% instead of the YData extrema alone
pcolor(X,Y,Z); shading flat
colormap(parula)
set(gca,'ydir','normal')
set(gca,'yscale','log')
xlabel('x'); ylabel('y')

I'm not really sure how that extends to your actual data, but just be careful to make sure your image is being interpolated as you expect it to be.

[Sneha Kandapal]

Thank you for the detailed explanation. I tried using the pocolor() and changed the code. It did not produce any image. I don't know how to flip that scale and still keep the image. I have attached the code and some input files.

% Prompt the user to select multiple data files
[dataFiles, dataDirectory] = uigetfile('*.txt', 'Select data files', 'MultiSelect', 'on');

% Check if the user canceled the file selection
if isequal(dataFiles, 0)
    disp('No files selected. Exiting...');
    return;
end

% Prompt the user to select an output directory
outputDirectory = uigetdir('Select an output directory', 'Specify Output Directory');

% Check if the user canceled the output directory selection
if outputDirectory == 0
    disp('No output directory selected. Exiting...');
    return;
end

% Initialize variables to store data
xData = [];
yData = [];

% Loop through selected data files
for i = 1:length(dataFiles)
    % Construct the full file path
    fullFilePath = fullfile(dataDirectory, dataFiles{i});
    
    % Read the data from the current file
    data = dlmread(fullFilePath);

    % Extract x and y data (1st and 4th columns)
    x = data(:, 1);
    y = data(:, 4);
    
    % Apply the formula to the y data
    y = (10.^((y - 3.51307) / 0.22845)) ./ (7.75E-5 * 0.2);
    
    % Append data to the storage variables
    xData = [xData; x];
    yData = [yData; y];
end

% Define the number of bins for the histogram
numXBins = 200; % Adjust as needed
numYBins = 100; % Adjust as needed

% Define the bin edges for the X and Y axes
xBinEdges = linspace(min(xData), max(xData), numXBins);
yBinEdges = logspace(log10(min(yData)), log10(max(yData)), numYBins);

% Create a 2D histogram color map with specific bin edges
h = histcounts2(xData, yData, xBinEdges, yBinEdges);

% Create a figure
figure;

% Define the grid matrices for pcolor
[X, Y] = meshgrid(xBinEdges, log10(flip(yBinEdges)));

% Create a pseudocolor plot using pcolor with reversed Y-axis scale
pcolor(X, Y, log(h)');
shading interp; % Use interpolated shading for smooth color transitions
colormap('jet'); % Use the 'jet' colormap for colorful visualization
colorbar; % Display the color bar for the Z-axis (point density)

% Customize the plot appearance
xlabel('X-Axis (Column 1)');
ylabel('Log(Y-Axis, reversed scale)'); % Y-axis is logarithmic and reversed
title('2D Pseudocolor Map of Multiple Data Files');

% Specify the output image file path
outputImageFilePath = fullfile(outputDirectory, 'pseudocolor_map.png');

% Save the pseudocolor plot as an image (e.g., PNG)
saveas(gcf, outputImageFilePath, 'png');

% Display the path where the image is saved
disp(['2D Pseudocolor Map saved to: ' outputImageFilePath]);

[DGM]

You can try either:

% Create a pseudocolor plot using pcolor with reversed Y-axis scale
pcolor(X, Y, imresize(log(h).',size(X)));
shading interp; % Use interpolated shading for smooth color transitions

or

imagesc(imresize(log(h).',size(X)),'xdata',X(:),'ydata',Y(:));
set(gca,'ydir','normal')

pcolor() will render NaNs differently than imagesc() does.

[Sneha Kandapal]

Thank you that worked. I just have to chnage Y to flip(Y).

pcolor(X, flip(Y), imresize(log(h).', size(X)));

Answer 2

[Sulaymon Eshkabilov]

Use 'reverse' option for the plot axis value direction, e.g.:

t=linspace(0,1.2, 1e3);
dt = t(2)-t(1);
Fs = 1/dt;
f1 = 5;
f2 = 25;
f3 = 100;
Z = sin(2*pi*t*f1)+0.75*sin(2*pi*t*f2)+0.5*cos(2*pi*t*f3)+0.25*randn(size(t));
[P,ff,tt] = pspectrum(Z,Fs,'spectrogram');
figure(1)
waterfall(ff,tt,P')
xlabel('Frequency (Hz)')
ylabel('Time (seconds)')
FIG = gca;
% X and Y axis values are in the reverse order
FIG.XDir = 'reverse';                % This reverses the values of x axis
FIG.YDir = 'reverse';                % This reverses the values of y axis
view([30 45])

% Compare to this one: all axis values are in default order given by MATLAB
% plot tools/fcns
figure(2)
waterfall(ff,tt,P')
xlabel('Frequency (Hz)')
ylabel('Time (seconds)')
view([30 45])

Answer 3

[Image Analyst]

Use either

axis ij

or

axis xy

until it's the way that you want it.

[Sneha Kandapal]

Thank you will try that.