Insert image into pie chart

Question

I am using python 3.5.2 I would like to make a pie chart with a nested png image. I have photos of some bulk products that I would like to put into pieces. For example, strawberries in one slice and raspberries in another. Very similar to the image http://www.python-course.eu/images/pie_chart_with_raspberries.png .

I can create images and even build images instead of points, as Matplotlib shows here : how to build images instead of points?

However, I could not find any approach to what I propose. I assume that this can be done manually in paint, but I tried to avoid this.

Answer 1

It is possible. We can start with a normal pie chart . Then we will need to get the images into the plot. This is done using plt.imread and using matplotlib.offsetbox.OffsetImage . We need to find good coordinates and zoom levels to place the image so that it completely overlaps with the corresponding pie wedge. Then the wedge path of the pie is used as the path of the image clip , so that only a part remains inside the wedge. Setting the zorder of an unfilled wedge to a large number ensures that borders are placed on top of the image. Thus, the wedges seem to be filled with an image.

 import matplotlib.pyplot as plt from matplotlib.patches import PathPatch from matplotlib.offsetbox import OffsetImage, AnnotationBbox total = [5,7,4] labels = ["Raspberries", "Blueberries", "Blackberries"] plt.title('Berries') plt.gca().axis("equal") wedges, texts = plt.pie(total, startangle=90, labels=labels, wedgeprops = { 'linewidth': 2, "edgecolor" :"k","fill":False, }) def img_to_pie( fn, wedge, xy, zoom=1, ax = None): if ax==None: ax=plt.gca() im = plt.imread(fn, format='png') path = wedge.get_path() patch = PathPatch(path, facecolor='none') ax.add_patch(patch) imagebox = OffsetImage(im, zoom=zoom, clip_path=patch, zorder=-10) ab = AnnotationBbox(imagebox, xy, xycoords='data', pad=0, frameon=False) ax.add_artist(ab) positions = [(-1,0.3),(0,-0.5),(0.5,0.5)] zooms = [0.4,0.4,0.4] for i in range(3): fn = "data/{}.png".format(labels[i].lower()) img_to_pie(fn, wedges[i], xy=positions[i], zoom=zooms[i] ) wedges[i].set_zorder(10) plt.show()