trying to find "corners" of puzzle pieces

corners.py

import numpy as np
import matplotlib.pyplot as plt

from skimage import feature
from skimage import filters
from skimage import io
from skimage import measure


# Load the image.
source = 'a.jpg'
piece = io.imread(source, as_grey=True)


# Find edges with the Canny filter.
low = 0.0
otsu = filters.threshold_otsu(piece, nbins=256)
sigma = 3
canny_edges = feature.canny(piece, sigma=sigma, low_threshold=low,
                            high_threshold=otsu)
canny_edges = np.logical_not(canny_edges)


# Get corners.
dist = 20
sensitivity = 0.05
window = 13
coords = feature.corner_peaks(
  feature.corner_harris(canny_edges, k=sensitivity), min_distance=dist)


# Display results.
fig, (ax1, ax2, ax3) = plt.subplots(nrows=1, ncols=3, figsize=(20, 8))

ax1.imshow(piece, cmap=plt.cm.gray)
ax1.axis('off')

ax2.imshow(canny_edges, cmap=plt.cm.gray)
ax2.axis('off')
ax2.set_title('Canny filter, $\sigma=%s$' % sigma, fontsize=20)

ax3.imshow(canny_edges, cmap=plt.cm.gray)
ax3.plot(coords[:, 1], coords[:, 0], '+r', markersize=12)
ax3.axis('off')
ax3.set_title('Harris corners', fontsize=20)

fig.subplots_adjust(wspace=0.02, hspace=0.02, top=0.9,
                    bottom=0.02, left=0.02, right=0.98)
plt.show()
fig.savefig('out.png')