Hi Siggi,

How about replacing

>>> a[find_boundaries(a)]=0

with

>>> s = nd.generate_binary_structure(2, 1)
>>> a[find_boundaries(a) * (nd.grey_erosion(a, footprint=s) != 0)] = 0

?

That will leave you with big objects that are separated with respect to connectivity=1, and objects that aren't touching will not be "reduced".

The output for your example is:

>>> print a.astype(int)
[[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 3 3 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 3 3 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 1 1 1 1 1 0 0 0 0 0 0]
 [0 0 0 0 1 1 1 1 1 0 0 0 0 0 0]
 [0 0 0 0 1 1 1 0 0 2 2 2 0 0 0]
 [0 0 0 0 1 1 1 0 2 2 2 2 0 0 0]
 [0 0 0 0 1 1 1 0 2 2 2 2 0 0 0]
 [0 0 0 0 0 0 0 2 2 2 2 2 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]]

Juan.

On Sat, Nov 24, 2012 at 3:13 AM, Sigmund <siggin@gmail.com> wrote:
Moin moin,

I'm using the find_boundaries function for separating labeled regions (separated by the watershed). Afterward I set the as True returned pixels to 0.  This works quit well but since my objects (labeled regions) can be small (couple pixels) and seldom stick together I would like to limit the find_boundaries to connected regions.
Is there a easy way to only find the boundaries not connected to 0. Or tell the watershed function to leave a space between separate objects?


Example:
import numpy as np
from scipy import ndimage as nd
from skimage.segmentation.boundaries import find_boundaries
a = np.zeros((15,15))
a[4:9,4:9]=1
a[6:10,7:12]=2
a[1:3,1:3]=3
print "a"
print a.astype(int)
a[find_boundaries(a)]=0
print a.astype(int)

Output:

[[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 3 3 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 3 3 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 1 1 1 1 1 0 0 0 0 0 0]
 [0 0 0 0 1 1 1 1 1 0 0 0 0 0 0]
 [0 0 0 0 1 1 1 2 2 2 2 2 0 0 0]
 [0 0 0 0 1 1 1 2 2 2 2 2 0 0 0]
 [0 0 0 0 1 1 1 2 2 2 2 2 0 0 0]
 [0 0 0 0 0 0 0 2 2 2 2 2 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]]
After find_boundaries:
[[0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 3 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 1 1 1 1 0 0 0 0 0 0]
 [0 0 0 0 0 1 1 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 1 1 0 2 2 2 2 0 0 0]
 [0 0 0 0 0 1 1 0 2 2 2 2 0 0 0]
 [0 0 0 0 0 0 0 0 2 2 2 2 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]
 [0 0 0 0 0 0 0 0 0 0 0 0 0 0 0]]

As you can see, the region 3 is almost gone.

Thanks
Siggi

--