Normalize Cut in Matrix Form
);    D(i,j) = 0
,
(
)
,
(
  
:
i
 
node
 
from
 
 costs
of
 
sum
 
 the
is
 
;
)
,
(
  
:
matrix
 cost
 the
is
 
,
j
i
i
i
c
j
i
j
j
i
=
=
ĺ
W
D
D
W
W
Can write normalized cut as:
•Solution given by “generalized” eigenvalue problem:
•
•Solved by converting to standard eigenvalue problem:
•
•
•optimal solution corresponds to second smallest eigenvector
•for more details, see
–J. Shi and J. Malik, Normalized Cuts and Image Segmentation, IEEE Conf. Computer Vision and Pattern Recognition(CVPR), 1997
–http://www.cs.washington.edu/education/courses/455/03wi/readings/Ncut.pdf