Automatic Representation and Segmentation of Video Sequences via a Novel Framework Based on the D-EVM and Kohonen Networks

Table 11

Size of the 3D-EVMs generated for every cluster associated with the Billiard Game Video Sequence under a SOM with 10 neurons, removal of the temporal component and unification of RGB components.

Class	Members	EVM size
0	82,064	13,552
1	340,045	45,358
2	250,389	9,012
3	180,390	26,104
4	66,439	8,896
5	49,017	6,072
6	15,538	2,726
7	672,032	37,044
8	73,421	11,524
9	75,657	12,088
Total	1,804,992	172,376