I want to prepare one selection of data from my high-quality PDF document which has no textual elements (just a plot), prepared originally by Matlab.
I do not want to give the whole picture for my collegues because it is too overwhelming.
#1 Tools in Matlab
I know this thread How can I read an image file that is stored in PDF format (much like reading a jpeg file with I = imread('image.jpg')? but I have got denying experiences from my colleagues and to my task PDF should be enough because my data is just a high-quality plot without textual elements.
Most relevant thread is this one How to extract data from pdf file in matlab?
Most attempts are based on extracting PDF to TXT, like How to Read PDF file in Matlab? about pdftotext.
I want now imcrop the PDF such that the output could be used in the time-series analysis of Mathematica here, but I did not find that the default imcrop tool of Matlab is supporting PDF, Crop an Image.
Some findings
Show and Save as PDF based on the answer. I do pdf = Import[filename.pdf]; Show[pdf[[1]], PlotRange -> {{50, 200}, {100, 300}}] and I see a good selected picture in Image viewer, but failure when exporting the picture back to Mathematica seeing the complete picture. Why? PlotRange does not crop but only put a white mask on the top of the picture which can separated etc in Mathematica.
Going from Show to ImageCrop based on this answer. Wrong approach, confusion with ImageTake.
Going from Show to ImageTake based on this answer.
The Show and ImageTake are not injective to each other because ImageTake has at least reversed order of parameters {ymin,ymax}, {xmin,xmax} according to the manual. However, I could not manage to select the correct selection by just reversing the parameters. Why?
Comments for Mathematica
It would be nice if the regions selected would correspond to each other.
Therefore, I would like to have some visual tool to select appropriate area from the figure.
I notice there occurs some aliasing when enlarging the original image.
It would be nice to know how Mathematica handles such cases with ImageTake.
How can you prepare imcrop of PDF image for the time-series toolbox of Mathematica?
I think this question is about image extraction.
However, I extended the question to the thread Better Colormap of Matlab and Image Extraction for Time-Series Toolbox of Mathematica? for Mathematica.
Mathematica will import your pdf as a graphic object which you can 'crop' using plotrange.
pdf = Import[filename.pdf];
Show[pdf[[1]], PlotRange -> {{50, 200}, {100, 300}}]
note the values are {{xmin,xmax},{ymin,ymax}} in "points"
You can also rasterize and then use ImageTake
ImageTake[Rasterize[pdf[[1]]], {10, 100}, {20, 100}]
here the values are {ymin,ymax} , {xmin,xmax} (note the reverse order )
Note the [[1]] here is effectively the page number. I'm pretty sure Import returns a list of pages even if the pdf is a single page.
If you want to actually extract plot data that's a whole other question. For that I'd suggest mathematica.stackexchange.com and provide an example file.
Related
I have Matlab code that produces an array of subplots that looks like this:
When I save this as an EMF file (normally a vector format) and insert in Powerpoint or view with Inkscape, then zoom in closely on the image, it looks pixelated, and clearly made up of a single encapsulated object as opposed to many small graphical objects:
This is very surprising to me, as I have for a long time been exporting Matlab plots as EMF and integrating those into Powerpoint slides, in order to build there more complex graphs that, as vector graphics, would print well at any size. These imported EMFs would also scale well and look smooth regardless of how much I zoomed into them in Powerpoint.
Can anyone guess why it is that on this occasion, this plot is not saved as a vector graphics but is instead (as it seems) rendered as raster? The code that produces this figure (based on the Matlab commands subplot, line, and scatter) is rather long/inelegant, but I can give details or simplify, if required to find a solution. Thanks!
Other similar threads on this site have not helped fix this.
I believe I have encountered this issue before. If it is the same as the issue I was facing, it is to do with the renderer being used to save the plot. Although it should automatically select the painters renderer when exporting to vector files, I have had instances where it used the openGL renderer instead, which results in bitmaps being used. I'm not entirely sure why this happens -- it might be the case that for particularly complicated figures, it reverts to openGL to avoid obnoxiously large or complicated vector files.
If you are using the print command, you can force it to use the painters algorithm as follows:
print('-painters',...)
if you're saving using File>Save As..., I believe setting the renderer for the figure should work:
set(figure_handle,'renderer','painters');
For explanation, per the MATLAB documentation:
-opengl' — OpenGL renderer. Use this renderer when saving bitmap images. OpenGL produces a bitmap image even with vector formats, which might limit the extent to which you can edit the image in other applications.
'-painters' — Painters renderer. Use this renderer when saving vector graphics files. If you save to a vector graphics file and if the figure RendererMode property is set to 'auto', then print automatically attempts to use the Painters renderer. If you want to ensure that your output format is a true vector graphics file, then specify the Painters renderer. For example:
print('-painters','-deps','myVectorFile')
If you do not specify the renderer, then print automatically uses the appropriate renderer to produce the output format requested. However, if you set the Renderer property for the figure, then print uses that renderer when generating output.
EDIT: Another option is to use the Copy Figure command (Edit > Copy Figure) -- this should copy the figure as an EMF file, and should obey the figure's renderer settings.
I am asking for a solution to the blurred image I get when I use copy figure option in Matlab. For example, when I plot 3d figure in matlab I get the following image (It is the result of printscreen!)..
However, when I use the option copy figure I get a blurred image as the following image
.
How can I avoid this as I need to get high resolution image using copy figure option. Thank you in advance for any help you may lend.
By default, the figure is copied as an Enhanced Metafile (a color graphics format) which works for basic plots such as bar plots, line plots, and other 2D plots but may yield undesirable results for more complex plots as in your case. This is because the rendering of the Metafile content is controlled my Microsoft Word and it may render things differently than MATLAB.
If you need a high resolution image, I would recommend using the print command where you can specify the desired resolution (using the -r option). Then you can import the resulting image into Word
print(gcf, 'myfile.png', '-dpng', '-r300')
Alternately, you can use export_fig from the File Exchange which better preserves the appearance
Another option would be to change the format used by Copy Figure to tell it to use a bitmap instead
Or you can call Copy Figure programmatically and specify a different format (either a bitmap or PDF)
print(gcf, '-clipboard', '-dpdf')
i know its a often discussed topic. But i haven't found any solution for me. I have some plots, which i need to import in Word 2013 afterwards the docx is saved as pdf.
The problem is:
If the plot gets saved (with the function print() ) as .eps or emf and imported in Word, the downsampled placeholder of the graphics look pretty poor and the graphics in pdf are no vector grapic anymore. They do get turned into a bitmap and by zooming in a bit, one can see the pixels. If i use Edit -> Copy Figure in the Matlab figure and paste it in Word, i get a nice graphic in Word as well as a nice vector based graphic in the pdf file.
Is there a way, to save a figure to a file and import it to Word and have the same quality like i have by copy and paste? What is the difference between these methods?
Thank you for your help.
Rafael
How the line smoothness in a contour plot can be improved for publications? For instance, the dotted lines look really bad, the continuous lines look as if their thickness varies. See below
Here's part of the code:
Vals = [0:5:200]; contourf(X,Y,W,Vals,'EdgeColor','k','LineWidth',1.2,'LineStyle',':');axis square;grid;hold on
Vals = [10:10:200]; contour(X,Y,W,Vals,'EdgeColor','k','LineWidth',1.2);
Vals = [20 : 20 : 200]; [C,h] = contour(X,Y,W,Vals,'Color','k','LineWidth',1.8);
clabel(C,h,'FontName','Palatino Linotype','FontAngle','italic','Fontsize',9,'Color','w')
print -djpeg -r300 filename
Thanks!
Saved as png doesn't help much... check the lines :/ See below:
Check the dotted lines now...
Here's saving as eps (-r1200)... it looks better
Exporting as vector graphics will definitely improve the image over what you see on your screen; I use LaTeX for publications and you can either export to eps for postscript output, and use epstopdf for PDF output, and embed these directly in your document; that would be the best solution.
Additionally, there are also a bunch of general utilities for making your plots look better for camera-ready publications, the most notable that comes to mind is exportfig, which has a load of features to help even with pixel graphics. These go above and beyond just generating smoother-looking images.
http://www.mathworks.us/matlabcentral/fileexchange/23629-exportfig
(copied from that page):
This function saves a figure or single axes to one or more vector and/or bitmap file formats, and/or outputs a rasterized version to the workspace, with the following properties:
Figure/axes reproduced as it appears on screen
Cropped borders (optional)
Embedded fonts (pdf only)
Improved line and grid line styles
Anti-aliased graphics (bitmap formats)
Render images at native resolution (optional for bitmap formats)
Transparent background supported (pdf, eps, png)
Semi-transparent patch objects supported (png only)
RGB, CMYK or grayscale output (CMYK only with pdf, eps, tiff)
Variable image compression, including lossless (pdf, eps, jpg)
Optionally append to file (pdf, tiff)
Vector formats: pdf, eps
Bitmap formats: png, tiff, jpg, bmp, export to workspace
This function is especially suited to exporting figures for use in publications and presentations, because of the high quality and portability of media produced.
Update: I see your example code now. Did you try changing -r300 to some really high value? More pixels per inch should make everything look smoother. For publication, crank it up really high, like -r1200.
Original:
One thing you can try is exporting the plot in some format that supports vector graphics. Matlab supports both PDF and EMF, so try one of those. Export using the saveas command or from the figure's "File -> Save as" menu item. After that, open or import the image file in some other application and hopefully it will look better.
Please add a new screenshot if you get a nicer image!
I have a 2D color-map plot created with imagesc and want to export it as a .eps file using
print -depsc.
The problem is that the "original" image data is from a rather small matrix (131 x 131). When I view the image in the matlab figure window, I can see all the individual pixels if I zoom a bit closer.
When I export to eps, however, there seems to be some interpolation or anti-aliasing going on, in that neighboring pixels get blurred/blended into each other. I don't get the problem if I export a high-resolution tiff, but that format is not an option (as demanded by a publisher).
How can I obtain an eps that preserves the pixely structure of my image without applying interpolation or anti-aliasing?
The blurring actually depends on the rendering software your viewer application or printer uses. To get good results all the time, make each pixel in your image an 8x8 block of pixels of the same color. The blurring then only affects the pixels at the edge of each block. 8x8 blocks are best as they compress without nasty artifacts using DCT compression (sometimes used in eps files).
Old question, but highly ranked in Google, so here is my answer:
Open the .eps-file with a text editor, search for "Interpolate" and change the following "true" to "false". Repeat that step for all Interpolate-statements.
It might also depend on the viewer you're using, but probably just because some viewers ignore the "Interpolate"s...
Had the same problem using plot2svg in Matlab and exporting from Inkscape to eps.