fix yen blank image bug

skimage/filter/tests/test_thresholding.py

@@ -43,10 +43,19 @@ class TestSimpleImage():
         assert threshold_yen(image) == 127
 
     def test_yen_binary(self):
-        image = np.zeros([2,256], dtype='uint8')
+        image = np.zeros([2,256], dtype=np.uint8)
         image[0] = 255
         assert threshold_yen(image) < 1
 
+    def test_yen_blank_zero(self):
+        image = np.zeros((5, 5), dtype=np.uint8)
+        assert threshold_yen(image) == 0
+
+    def test_yen_blank_max(self):
+        image = np.empty((5, 5), dtype=np.uint8)
+        image.fill(255)
+        assert threshold_yen(image) == 255
+
     def test_threshold_adaptive_generic(self):
         def func(arr):
             return arr.sum() / arr.shape[0]
@@ -124,5 +133,10 @@ def test_yen_coins_image_as_float():
     assert 0.43 < threshold_yen(coins) < 0.44
 
 
+def test_yen_camera_image():
+    camera = skimage.img_as_ubyte(data.camera())
+    assert 197 < threshold_yen(camera) < 199
+
+
 if __name__ == '__main__':
     np.testing.run_module_suite()

skimage/filter/thresholding.py

@@ -172,15 +172,16 @@ def threshold_yen(image, nbins=256):
     >>> binary = image <= thresh
     """
     hist, bin_centers = histogram(image, nbins)
-    norm_histo = hist.astype(float) / hist.sum() # Probability mass function
-    P1 = np.cumsum(norm_histo) # Cumulative normalized histogram
+    if bin_centers.size == 1:
+        return bin_centers[0]
+
+    norm_histo = hist.astype(float) / hist.sum()  # Probability mass function
+    P1 = np.cumsum(norm_histo)  # Cumulative normalized histogram
     P1_sq = np.cumsum(norm_histo ** 2)
     # Get cumsum calculated from end of squared array:
     P2_sq = np.cumsum(norm_histo[::-1] ** 2)[::-1]
     # P2_sq indexes is shifted +1. I assume, with P1[:-1] it's help avoid '-inf'
     # in crit. ImageJ Yen implementation replaces those values by zero.
-    crit = np.log(((P1_sq[:-1] * P2_sq[1:]) ** -1) * \
+    crit = np.log(((P1_sq[:-1] * P2_sq[1:]) ** -1) *
                   (P1[:-1] * (1.0 - P1[:-1])) ** 2)
-    max_crit = np.argmax(crit)
-    threshold = bin_centers[:-1][max_crit]
-    return threshold
+    return bin_centers[crit.argmax()]