Re: [Matplotlib-users] Simple scatter plot over an image

[__ <red...@gm...>]

Excellent. Thank you very much!

On 6/10/07, Jake Emerson <jak...@on...> wrote:
>
>  The python imaging library is pretty good for this kind of thing.
>
> http://www.pythonware.com/library/pil/handbook/
>
> Here's an (untested) example. Hope it helps.
>
> Jake
>
>
> #!/usr/bin/env python
>
> from pylab import scatter, save
> import Image
>
> #get the background image, and find out how big it is
> im_bg = Image.open("background.png")
> bg_width, bg_height = im.size
>
>  #make a white canvas on which to paste the background image and the
> scatter plot
> #this will allow you to, say, have the x- and y-axis values fall outside
> of the background image's limits
> im_canvas = Image.new("RGBA", (bg_width+60, bg_height+60), (255, 255, 255,
> 0))
>
> #create the scatter plot from x and y data with matplotlib
> scatter(x, y, s=sizes, alpha=0.75)
>
> #save the scatter plot, and then retrieve it for use in PIL, convert to
> RGBA so that alpha levels will work
> #there is probably a better way to do this with gcf() or gci()...
> savefig("scatter_plot.png")
> im_scatter = Image.open("scatter_plot.png").convert("RGBA")
>
> #resize the scatter image to make it fit nice
> im_scatter.resize((bg_width+10, bg_height+10))
>
> #bring all of the images together by pasting them onto the white canvas,
> use the overlayed image as the mask (third element)
> im_canvas.paste(im_bg, (30, 30), im_bg)                   #play around
> with the paste locations (30, 30)
> im_canvas.paste(im_scatter, (10, 30), im_scatter)      #these won't be
> perfect the first time (10, 30)
>
> #save it
> im_canvas.save("combo_image.png")
>
>
>
>
>  ------------------------------
>  *From:* __ [mailto:red...@gm...]
> *Sent:* Sunday, June 10, 2007 5:22 PM
> *To:* mat...@li...
> *Subject:* [Matplotlib-users] Simple scatter plot over an image
>
> Hello,
>
> I'm trying to plot a simple list of x/y coords over an image (.png). I can
> show the image, or plot the data, but cannot find a way to layer one over
> the other. I would greatly appreciate someone pointing me in the right
> direction. Thanks.
>