Small fixes on the Harris corner detector - documentation, method renaming etc

CONTRIBUTORS.txt

@@ -83,6 +83,7 @@
 
 - Nelle Varoquaux
   Renaming of the package to ``skimage``.
+  Harris corner detector
 
 - W. Randolph Franklin
   Point in polygon test.

doc/examples/plot_harris.py

@@ -3,29 +3,28 @@
 Harris Corner detector
 ===============================================================================
 
-The Harris corner filter detects interest points using edge detection in many
-direction.
+The Harris corner filter detects interest points using edge detection in
+multiple direction.
 """
 
 from matplotlib import pyplot as plt
-from matplotlib import cm
 
-from skimage import data
-from skimage.filter import harris_corner_detector
+from skimage import data, img_as_float
+from skimage.filter import harris
 
 
 def plot_harris_points(image, filtered_coords):
     """ plots corners found in image"""
 
-    plt.subplot(111)
-    plt.imshow(image, cmap=cm.gray)
+    plt.plot()
+    plt.imshow(image)
     plt.plot([p[1] for p in filtered_coords],
-                [p[0] for p in filtered_coords],
-                'b.')
+             [p[0] for p in filtered_coords],
+             'b.')
     plt.axis('off')
     plt.show()
 
 
-im = data.lena().astype(float)
-filtered_coords = harris_corner_detector(im, 6)
+im = img_as_float(data.lena())
+filtered_coords = harris(im, 6)
 plot_harris_points(im, filtered_coords)

skimage/filter/__init__.py

@@ -5,4 +5,4 @@ from edges import sobel, hsobel, vsobel, hprewitt, vprewitt, prewitt
 from tv_denoise import tv_denoise
 from rank_order import rank_order
 from thresholding import threshold_otsu
-from harris import harris_corner_detector
+from harris import harris

skimage/filter/harris.py

@@ -10,18 +10,20 @@ from scipy import ndimage
 
 def _compute_harris_response(image, eps=1e-6):
     """Compute the Harris corner detector response function
-        for each pixel in the image
+    for each pixel in the image
 
-    Params
-    -------
-    image: ndarray
+    Parameters
+    ----------
+    image: ndarray of floats
+        input image
 
-    eps: float, optional, default: 1e-6
-    normalisation factor
+    eps: float, optional
+        normalisation factor
 
     Returns
     --------
-    ndarray
+    features: (M, 2) ndarray
+        Harris image response
     """
     if len(image.shape) == 3:
         image = image.mean(axis=2)
@@ -42,8 +44,10 @@ def _compute_harris_response(image, eps=1e-6):
 
     # Non maximum filter of size 3
     harris_max = ndimage.maximum_filter(harris, 3, mode='constant')
-    harris *= harris == harris_max
-    # Remove the image corners
+    mask = (harris == harris_max)
+    harris *= mask
+
+    # Remove the image borders
     harris[:3] = 0
     harris[-3:] = 0
     harris[:, :3] = 0
@@ -52,23 +56,26 @@ def _compute_harris_response(image, eps=1e-6):
     return harris
 
 
-def harris_corner_detector(image, min_distance=10, threshold=0.1, eps=1e-6):
+def harris(image, min_distance=10, threshold=0.1, eps=1e-6):
     """Return corners from a Harris response image
 
-    params
-    -------
-    harrisim: ndarray of floats
+    Parameters
+    ----------
+    image: ndarray of floats
+        Input image
 
-    min_distance: int, optional, default: 10
-        minimum number of pixels separating corners and image boundary
+    min_distance: int, optional
+        minimum number of pixels separating interest points and image boundary
 
-    threshold: float, optional, default: 0.1
+    threshold: float, optional
+        relative threshold impacting the number of interest points.
 
-    eps: float, optional, default: 1e-6
+    eps: float, optional
+        Normalisation factor
 
     returns:
     --------
-    array: coordinates
+    array: coordinates of interest points
     """
     harrisim = _compute_harris_response(image, eps=eps)
     corner_threshold = np.max(harrisim.ravel()) * threshold
@@ -78,8 +85,9 @@ def harris_corner_detector(image, min_distance=10, threshold=0.1, eps=1e-6):
 
     # get coordinates of candidates
     candidates = harrisim_t.nonzero()
-    coords = [(candidates[0][c], candidates[1][c]) for c
-               in range(len(candidates[0]))]
+    coords = np.concatenate((candidates[0].reshape((len(candidates[0]), 1)),
+                             candidates[1].reshape((len(candidates[0]), 1))),
+                            axis=1)
 
     # ...and their values
     candidate_values = [harrisim[c[0]][c[1]] for c in coords]

skimage/filter/tests/test_harris.py

@@ -1,23 +1,28 @@
 import numpy as np
 
-import unittest
-
-from skimage.filter import harris_corner_detector
+from skimage.filter import harris
+from skimage import img_as_float
 
 
-class TestHarris(unittest.TestCase):
-
+class TestHarris():
     def test_square_image(self):
         im = np.zeros((50, 50)).astype(float)
         im[:25, :25] = 1.
-        results = harris_corner_detector(im)
-        self.assertTrue(results.any())
-        self.assertTrue(len(results) == 1)
+        results = harris(im)
+        assert results.any()
+        assert len(results) == 1
 
     def test_noisy_square_image(self):
         im = np.zeros((50, 50)).astype(float)
         im[:25, :25] = 1.
         im = im + np.random.uniform(size=im.shape) * .5
-        results = harris_corner_detector(im)
+        results = harris(im)
         assert results.any()
         assert len(results) == 1
+
+    def test_squared_dot(self):
+        im = np.zeros((50, 50))
+        im[4:8, 4:8] = 1
+        im = img_as_float(im)
+        results = harris(im)
+        assert results == np.array([6, 6])