Scale image in matplotlib without changing the axis

Question

I have a GUI that displays a plot. I want to fit that plot to an existing image. I displayed the image under the plot using:

myaxe.plot(...)
myaxeimage = mya         


        

Answer 1

There is a watermark example distributed with matplotlib that is sort of similar. Starting from that code, we can modify as follows:

Use ax.imshow to plot the image first. I do this because the extent parameter affects the final extent of ax. Since we want the final extent to be governed by the plt.plot(...), let's put it last.

myaximage = ax.imshow(im, aspect='auto', extent=(1,15,0.3,0.7), alpha=0.5, origin='upper', zorder=-1)

Instead of extent=myaxe.axis(), use extent to control the position and size of the image. extent=(1,15,0.3,0.7) places the image in the rectangle with (1, 0.3) as the bottom left corner and (15, 0.7) as the top right corner.

With origin='upper', the [0,0] index of the array im is placed at the upper left corner of the extent. With origin='lower' it would have been placed at the lower left corner.

---

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.cbook as cbook
import matplotlib.image as image
np.random.seed(1)
datafile = cbook.get_sample_data('logo2.png', asfileobj=False)
im = image.imread(datafile)
fig, ax= plt.subplots()

myaximage = ax.imshow(im, aspect='auto', extent=(1,15,0.3,0.7), alpha=0.5, zorder=-1)
ax.plot(np.random.rand(20), '-o', ms=20, lw=2, alpha=1.0, mfc='orange')
ax.grid()
plt.show()

---

If you want to expand the image and clip it to the extent of the plot, you might need to use ax.set_xlim and ax.set_ylim as well:

myaximage = ax.imshow(im, aspect='auto', extent=(-1,25,0.3,0.7), alpha=0.5, zorder=-1,
                      origin='upper')

ax.plot(np.random.rand(20), '-o', ms=20, lw=2, alpha=1.0, mfc='orange')
ax.set_xlim(0,20)
ax.set_ylim(0,1)

---

Or, for more control, you can clip the image to an arbitrary path by using myaximage.set_clip_path:

import numpy as np
import matplotlib.pyplot as plt
import matplotlib.cbook as cbook
import matplotlib.image as image
import matplotlib.patches as patches
np.random.seed(1)
datafile = cbook.get_sample_data('logo2.png', asfileobj=False)
im = image.imread(datafile)
fig, ax= plt.subplots()

myaximage = ax.imshow(im, aspect='auto', extent=(-5,25,0.3,0.7), 
                      alpha=0.5, origin='upper',
                      zorder=-2)
# patch = patches.Circle((300,300), radius=100)
patch = patches.Polygon([[5, 0.4], [15, 0.4], [15, 0.6], [5, 0.6]], closed=True,
                        transform=ax.transData)
myaximage.set_clip_path(patch)
ax.plot(np.random.rand(20), '-o', ms=20, lw=2, alpha=1.0, mfc='orange', 
        zorder=-1)

ax.set_xlim(0, 20)
ax.set_ylim(0, 1)

plt.show()

Answer 2

Finally, I followed tcaswell suggestion and used 2 different axes. This way, I simply have to play with set_xlim() and set_ylim() of my image axes to change the origin and/or the zooming factor of my image. I order to get the image below my plot, without hiding it with the frame of the plot, I removed the frame of the plot and used the frame of the image axes instead. I also hidden the ticks from the image axes.

from matplotlib import pyplot

f = pyplot.figure()
a = f.add_subplot(111, frameon=False) # Remove frame
a.plot(...)

myimg = pyplot.imread(...)
imgaxes = f.add_axes(a.get_position(), # new axes with same position
    label='image', # label to ensure imgaxes is different from a
    zorder=-1, # put image below the plot
    xticks=[], yticks=[]) # remove the ticks
img = imgaxes.imshow(myimg, aspect='auto') # ensure image takes all the place

# now, to modify things
img.set_alpha(...)
imgaxes.set_xlim((x1, x2)) # x1 and x2 must be calculated from
                           # image size, origin, and zoom factor