Add note about dtype overflow

skimage/filter/rank/_percentile.py

@@ -323,7 +323,8 @@ def sum_percentile(image, selem, out=None, mask=None, shift_x=False,
 
     Only greyvalues between percentiles [p0, p1] are considered in the filter.
 
-    Note that the result is truncated to 8 or 16 bit.
+    Note that the sum may overflow depending on the data type of the input
+    array.
 
     Parameters
     ----------

skimage/filter/rank/bilateral.py

@@ -174,6 +174,9 @@ def sum_bilateral(image, selem, out=None, mask=None, shift_x=False,
     Only pixels belonging to the structuring element AND having a greylevel
     inside this interval are summed.
 
+    Note that the sum may overflow depending on the data type of the input
+    array.
+
     Parameters
     ----------
     image : 2-D array (uint8, uint16)

skimage/filter/rank/generic.py

@@ -578,6 +578,9 @@ def pop(image, selem, out=None, mask=None, shift_x=False, shift_y=False):
 def sum(image, selem, out=None, mask=None, shift_x=False, shift_y=False):
     """Return the local sum of pixels.
 
+    Note that the sum may overflow depending on the data type of the input
+    array.
+
     Parameters
     ----------
     image : 2-D array (uint8, uint16)