Why is the entire image copied instead of the axes only?

What I have is a plot showing the area of connected components. What I want to do is to further work on the plot figure such as clean it up a bit or ``imcomplement`` it etc. and then be able to apply the axes from the original plot to this image and be able to extract the ``ylabel``.

Let me explain the above issue with my code and some examples.

This is the plot I have, the y-axis represents the object areas. This is the important axis that I want to transfer to the new image.

Since I am interested in the axes only I copy that using

    h = findobj(gcf,'type','axes');

So that I can work with the figure without the axes and borders interfering I save it without these attributes

    set(gca, 'visible', 'off'); % Hide the axis and borders
    hgexport(gcf, 'plot1.jpg', hgexport('factorystyle'), 'Format', 'jpeg');

This is what I get:

So far so good.

Now comes the processing or in other words changing the plot to my needs.

    plot_img = rgb2gray(imread('plot1.jpg'));
    img_bw_plot = im2bw(plot_img, graythresh(plot_img));

    [rows cols] = size(plot_img);

    new = zeros(size(plot_img));

    for i = 1: rows
        for j = 1: cols
            if (img_bw_plot(i,j) == 0)
                new(i, 1:10) = 255;
            end
        end
    end

    f = figure;

    imshow(new);
    copyobj(h,f)

This produces a weird overlapped image where instead of copying only the axes, the entire image is copied on top of the ``new``. The ``datacursormode`` also fails to work beyond the over lapping image.

Answer 1

[Image Analyst]

I really don't understand what you want to do. What or where is this "ylabel" that you want to extract? Do you mean the tick labels on the y axis? And I don't know why you want to treat a plot of markers (diamonds?) as if it were an image. And I don't know what you are hoping to do with the "new" matrix.

Why don't we take a step back and start with the original x,y data you plotted. You say you have a "plot showing the area of connected components." I guess x is the blob number, and y is the area of the blob. Now, what do you want to do with that?

[Faraz]

Hello ImageAnalyst,

I'm sorry I wasnt clear in explaining my problem. So let me start by saying what I actually want to do or in other words what my objective is.

Yes you are right by Ylabel I mean the tick labels on the y axis, and these represent the areas of the blob as you clearly guessed.

I want to extract the area of each blob via command line, as in make Matlab automatically pick up the area of the blob, for example the area of the top blob is roughly between 2700 and 4000.

This is what I WANT to achieve and this is my approach or attempt at it.

I found out that I can get all the y'axis values of the plot using this command:

>> h = plot((1:5).^2); %// example plot
>> get(h,'YData')

My approach was to use this command and find the "jumps" in the answer, for example if one blob finishes at 2500 and the other starts at 2700 I would know where one blob ends and where the other starts. And that is why I treated the plot as an image. Since it is in this zoomed out state I can see the blobs and at this zoom level the blobs actually represent the areas of the connected components. When zoomed in they look like this:

So I saved it as an image using

hgexport(gcf, 'plot1.jpg', hgexport('factorystyle'), 'Format', 'jpeg');

Then the concept of the new image is to purely translate the blobs into rectangles so that the jumps can be easily detected, let me comment that part of the code:

   plot_img = rgb2gray(imread('plot.jpg')); %read in the saved plot
  img_bw_plot = im2bw(plot_img, graythresh(plot_img)); %convert to logical
  [rows cols] = size(plot_img); %get rows and colums
  new = zeros(size(plot_img)); %make a new image, which is the same 
                              %size as of the original saved plot
      for i = 1: rows
          for j = 1: cols
              if (img_bw_plot(i,j) == 0) % If at this pixel a plot point
                                         %is detected then
                  new(i, 1:10) = 255;  %at same row of new make the first
                                      %ten colums white
              end
          end
      end

This would produce this image

Now finally in order to get the actual ylabels I wanted to copy the axes (Ylabel?) from the original plot onto this new image and in this way get all the Ylables, find the "jumps" and have my area of blobs.

This sir, is my problem and my approach to it.

I would really appreciate your expert opinion on it and if this approach is flawed I would appreciate if you could guide me in the right path as to how can I get the Ylabels from the actual plot?

Thank you

[Image Analyst]

I still don't know why you're messing with the plot. You have blob areas from the connected component (regions) so you did something like this

measurements = regionprops(labeledImage, 'Area');
allAreas = [measurements.Area];
plot(allAreas);

So you already have the areas even before you plot them. To try to get the areas by doing things like extracting YData from the plot or reading the ytick labels that it automatically created when it plotted your areas is, well, I just see no reason at all for it. Again, you have the areas already so why do all that other stuff?

[Faraz]

Yes, you are right, I already have the areas but its not that simple. The code you provided to plot all areas produces only this:

Not blobs but points. Their areas in this state are of no use to me. It is only after further processing (such as plotting the areas of the connected components at varying threshold levels with hold on) do they create blobs. These blobs represent areas that I need further down in my code.

So I am forced to do all this other stuff because only after the blobs are made I must extract their areas. that is read their y-labels.

So can you please help me out in extracting the y-label values of the blobs from the plot? Is my approach flawed? Can it be accomplished using Matlab?

Thank you

[Image Analyst]

I don't understand it enough to say if it's flawed or not, but here's code to extract the y tick labels:

plot(1:100);
theYLabels = get(gca, 'yticklabel')

It seems like you want to plot sets of asterisks for different settings and accumulate them on the plot so that they overlap and form blobs and then get an image of the plot area and treat the merged asterisks as blobs. If that's what you're intending, then yes, that's the wrong approach. You need to work with the area arrays directly, not from an image of them plotted.

[Faraz]

hmm. This just gave me the yticklabels, I just got a char array like this

0   
1000
2000
3000
4000
5000
6000

But I think this is what I asked for in the code you gave. This is not what I want.

I would really appreciate a way to solve this issue. I have already explained above but will do so again in summary.

 - I have a plot of points, these are, when zoomed out, blobs that represent objects.

 - How do I get the starting and ending y-label value of each blob from the plot from the command line?

I found out that this

 >> h = plot((1:5).^2); %// example plot
 >> get(h,'YData')

would give me the actual y-label values of the data, but since the data is only useful when zoomed out this does not give me the actual y-values. The results it gives back appears randomized.

I further found out that I can extract the points I need using "getdatamode" and transfer the points to command line, this works well but its not automatic. The system waits for my input

dcm_obj = datacursormode(figure(1));
set(dcm_obj,'DisplayStyle','datatip',...
'SnapToDataVertex','off','Enable','on');
pause();
c_info = getCursorInfo(dcm_obj);
pos = [c_info.Position];
new_pos = pos(2:2:end);

So I basically require the above code to work automatically.

Can you, with your expertise in image processing in Matlab suggest me how I can tackle this problem?

Thanks a lot

[Image Analyst]

Instead of plotting those points with plot() or scatter() or whatever you used, why don't you just set pixels in an array (an image) and just use regular image processing methods?

[Faraz]

Yes, I will look into it. That looks to be the best option. Since the area are plotted on the y-axis, I used this code to trasfer that to a vector where the area is the vector index. I am setting that pixel to white.

area = [stats.Area];
     normalized_area = area/norm(area);
     figure(1), plot(5, area(:),'.');
     axis([0 300 0 6000]);
     for k = 1: length(area)
         test(area(k)) = 255;
     end

Now what I have to do is to search for continuous "255" in the vector and that will be my blob.

Is this what you had in mind? with writing it directly to an array or is there a better way to do it?