Notation for k-means algorithm

N

number of pixels in an image

K

number of classes

k

index for the current class

i

index for the current pixel

x_i

data in the i'th sample. x_i is usually a vector.

k(i)

class to which x_i belongs

I_k

set of i's which belong to class k

n_k

number of elements in I_k

c_k = Sigma(i in I) (x_i)/n_k

the center of the K'th class

c = Sigma(k) n_k c_k/N = Sigma(i) c_k(i)/N

the center of the entire data set

Note that:

   N = Sigma(k) nk

Notation taken from [1]

This page is maintained by Prof. Miriam Leeser
Last updated September 16, 1999.
Email: mel@ece.neu.edu