MATLAB: Position of scatter points problem

position of scatter points

I have a set of scatter points (A), with unit of 'meter'. Then I plot them using:

f = figure(1)
scatter(A(:,1), A(:,2), 3, 'k', 'filled')

but what I noticed is that, the default Units of a figure is 'Pixel', so my question is: is there anyway to know what are the positions of these scatter points in figure?

Thanks!

Best Answer

https://www.mathworks.com/help/images/ref/axes2pix.html can give you pixels relative to the axes position. You can

ax = gca;
oldunits = get(ax, 'Units');
set(ax, 'Units', 'pixels');
axpos = get(ax, 'Position');
set(ax, 'Units', oldunits);

and now axpos will be in pixels relative to the figure, to which you could add the results of axes2pix.

I seem to remember there being a more direct internal routine, but I do not remember the name of it.

Related Solutions

MATLAB: How to calculate how many plot tick would exist if the axis completely filled the figure

Given an axis within a figure, compute what the x and y axis limits would be if the axis extended to the edges of the figure.

Demo: See inline comments that explain each step. The variables newXLim and newYLim contain the extended axis limits.

It uses randomly placed axes with randomly set axis limits. After computing the expanded axis limits a second axis is created set to fill the figure and contains the extended axis limits. This axis is used to plot lines that should align with the original axis limits to confirm successful conversion. The axis limits to the exapanded axes are shown in the upper, left corner.

% Create figure with randomly placed axes
% and random data spans (axis limits) to 
% test for maximum flexibility
fig = clf(); 
ax = axes(fig,'Units','Normalize',...
    'Position', [rand(1,2)*.4+.05, rand(1,2)*.2+.35], ...
    'LineWidth',3);
xlim(ax,sort(rand(1,2)*100))
ylim(ax,sort(rand(1,2)*100))
box(ax,'on')
% Set units to pixels
ax.Units = 'pixels'; 
fig.Units = 'pixels';
% Get locations and sizes
axPos = ax.InnerPosition;   % inner positions!
figPos = fig.Position;      % we only need width and height
% Compute margins between fig and ax edges 
% in pixels
axMarg = [axPos(1:2),figPos(3:4) - (axPos(1:2)+axPos(3:4))]; %[left, bottom, right, top]
% Now we know how far to extend the axes to each edge
% of the figure in pixel units.  
% Convert from pixel units to data units.  
xl = xlim(ax); 
yl = ylim(ax); 
xyFactor(1) = range(xl) / axPos(3);
xyFactor(2) = range(yl) / axPos(4);
xyMargDataUnits = axMarg.*xyFactor([1,2,1,2]); 
% Now we know the distance between each axis edge
% and figure edge in data units.  We just need to 
% extrapolate from the axis limits
newXLim = [xl(1)-xyMargDataUnits(1), xl(2)+xyMargDataUnits(3)];
newYLim = [yl(1)-xyMargDataUnits(2), yl(2)+xyMargDataUnits(4)];
% Set axis units back to normalized so if the figure 
% is resized, the results will not change (important!). 
ax.Units = 'normalize';
% Now let's test it by adding an underlying axis with these limits
ax2 = axes('Units','Normalize','InnerPosition',[0 0 1 1]);
box(ax2,'on')         % added after creating GIF image. 

ax2.LineWidth = 1.5;  % added after creating GIF image. 
uistack(ax2,'bottom')
xlim(ax2, newXLim)
ylim(ax2, newYLim)
% Use reference lines to confirm that the axis limits of 
% th original axes align with those values on the expanded axis.
xline(xl(1),'b--', 'LineWidth', 1.5)
xline(xl(2),'b--', 'LineWidth', 1.5)
yline(yl(1),'b--', 'LineWidth',1.5)
yline(yl(2),'b--', 'LineWidth',1.5)
% Show expanded axis limits
text(ax2,min(xlim(ax2)),max(ylim(ax2)),...
    sprintf('xlim: [%.1f, %.1f]\nylim: [%.1f, %.1f]', ...
    newXLim(1),newXLim(2),newYLim(1),newYLim(2)), ...
    'VerticalAlignment','Top','HorizontalAlignment','Left',...
    'FontSize', 12, 'FontWeight','bold','fontName','fixedwidth',...
    'BackgroundColor', 'w')

The image below shows several iterations of this demo.

MATLAB: Outputting text on subplot with mixed normalised/normal coordinates

There might be a different way, this is what comes to mind:

label_offset_pixels = 15;   %space in pixels between axes box and text to write

ax = gca;
old_units = get(ax, 'Units');
set(ax, 'Units', 'pixels');
ax_pos = get(ax, 'Position');
set(ax, 'Units', old_units);
label_x_pos_pixels = ax_pos(1) + ax_pos(3) + label_offset_pixels;
probe_text = text(label_x_pos_pixels, 1, 'test', 'Units', 'pixels');
set(probe_text, 'Units', 'data');
probe_pos_data = get(probe_text, 'Position');
label_x_pos_data = probe_pos_data(1);
delete(probe_text);
text(label_x_pos_data, y20, '20th pct');
text(label_x_pos_data, y80, '80th pct');

What this does is find the right boundary of the axes in pixels, add a margin, convert that to data units, and use the data units for the text position.

There is a considerably shorter way to find the normalized desired y coordinate, using the axes YLim properties, but that does not give you a fixed spacing between the box and the text. Though now that I think of it, you could do that, convert the result to Pixels, add the offset... yes, I guess that would be shorter in spirit

label_offset_pixels = 15;   %space in pixels between axes box and text to write
ax = gca;
ax_ylim = get(ax, 'Ylim');
normalized_y20_80 = ([y20, y80]-ax_ylim(1))./(ax_ylim(2)-ax_ylim(1));
pct_text = text([1, 1], normalized_y20_80, {'20th pct', '80th pct'});   %data units is default
set(pct_text, 'Units', 'pixel');
pct_text_pos = get(pct_text, 'Position');
set(pct_text, {'Position'}, cellfun(@(xyz) xyz+[label_offset_pixels 0 0], pct_text_pos,'Uniform', 0));
set(pct_text, 'Units', 'data');

Best Answer

Related Solutions

MATLAB: How to calculate how many plot tick would exist if the axis completely filled the figure

MATLAB: Outputting text on subplot with mixed normalised/normal coordinates

Related Question