1: Initialize cluster centers Ck
= [Lk, ak, bk, xk, yk, zk]T by sampling pixels at regular grid steps. |
2: Choose cluster centers in an n * n neighborhood, to the lowest gradient position. |
3: Repeat
|
4: for each cluster center Ck in zk
do
|
5: Assign the best matching pixels from an 2S * 2S square neighborhood around the cluster center according to the distance measure. |
6: end for
|
7: Compute new cluster centers and residual error E{distance between previous centers and recomputed centers}. |
8: until E < = threshold. Save cluster centers Ck and transmit Ck to the next image zk+1. |
9: until all image sequences are segmented. |
10: Enforce connectivity. |