How to get the images corresponding to each shape of the tree ?

[nboutry]

Hi all,

I am trying to get the images corresponding to the shapes of an image (using the tree of shapes).

I do not speak about the nodes of the tree, but about an image containing ones where each shape is defined, and zeros elsewhere.

If someone can help me, I will be very grateful.

Thanks for your help.
Best,
Nicolas

[PerretB]

Hello,

the following example demonstrates how to get a binary mask giving all the vertices of the base graph (pixels) that do (not) belong to a node of a hierarchy:

import higra as hg
import numpy as np
import matplotlib.pyplot as plt
from skimage import data

im = data.coins()
plt.imshow(im)

# construct the tree of shapes of im
tree, altitudes = hg.component_tree_tree_of_shapes_image2d(im)

# perform an area filter to remove small components of the tree
area = hg.attribute_area(tree)
tree, _ = hg.simplify_tree(tree, area < 100)

# construct a vector that will be true for any node expect the one we want
deleted = np.ones(tree.num_vertices(), dtype=np.bool)
# arbitrarily select the 50th node from the root
deleted[tree.root()-50] = False

# reconstruct the selected node (every node except the selected one)
pixels = hg.reconstruct_leaf_data(tree, deleted, deleted)

plt.figure()
plt.imshow(pixels)