1"""
2=================
3Template Matching
4=================
5
6We use template matching to identify the occurrence of an image patch
7(in this case, a sub-image centered on a single coin). Here, we
8return a single match (the exact same coin), so the maximum value in the
9``match_template`` result corresponds to the coin location. The other coins
10look similar, and thus have local maxima; if you expect multiple matches, you
11should use a proper peak-finding function.
12
13The ``match_template`` function uses fast, normalized cross-correlation [1]_
14to find instances of the template in the image. Note that the peaks in the
15output of ``match_template`` correspond to the origin (i.e. top-left corner) of
16the template.
17
18.. [1] J. P. Lewis, "Fast Normalized Cross-Correlation", Industrial Light and
19       Magic.
20
21"""
22import numpy as np
23import matplotlib.pyplot as plt
24
25from skimage import data
26from skimage.feature import match_template
27
28
29image = data.coins()
30coin = image[170:220, 75:130]
31
32result = match_template(image, coin)
33ij = np.unravel_index(np.argmax(result), result.shape)
34x, y = ij[::-1]
35
36fig = plt.figure(figsize=(8, 3))
37ax1 = plt.subplot(1, 3, 1)
38ax2 = plt.subplot(1, 3, 2)
39ax3 = plt.subplot(1, 3, 3, sharex=ax2, sharey=ax2)
40
41ax1.imshow(coin, cmap=plt.cm.gray)
42ax1.set_axis_off()
43ax1.set_title('template')
44
45ax2.imshow(image, cmap=plt.cm.gray)
46ax2.set_axis_off()
47ax2.set_title('image')
48# highlight matched region
49hcoin, wcoin = coin.shape
50rect = plt.Rectangle((x, y), wcoin, hcoin, edgecolor='r', facecolor='none')
51ax2.add_patch(rect)
52
53ax3.imshow(result)
54ax3.set_axis_off()
55ax3.set_title('`match_template`\nresult')
56# highlight matched region
57ax3.autoscale(False)
58ax3.plot(x, y, 'o', markeredgecolor='r', markerfacecolor='none', markersize=10)
59
60plt.show()
61