Improved Unsupervised Color Segmentation Using a Modified Color Model and a Bagging Procedure in -Means++ Algorithm

Table 8

Segmentation performance results for image “d” using GMM.

GMM	Cluster 1	Cluster 2	Cluster 3	Cluster 4	Metrics
					Best BAcc.: 0.96009; Mean BAcc: 0.88859; Mean sen.: 0.80937; Mean spe.: 0.92209; Avg. time: 3.97900
n					Best BAcc.: 0.94253; Mean BAcc: 0.75979; Mean sen.: 0.05137; Mean spe.: 0.98592; Avg. time: 5.96510
					Best BAcc.: 0.96159; Mean BAcc: 0.89942; Mean sen.: 0.82844; Mean spe.: 0.92873; Avg. time: 3.47550
					Best BAcc.: 0.97867; Mean BAcc: 0.82215; Mean sen.: 0.85910; Mean spe.: 0.81874; Avg. time: 1.8572
YCrCb					Best BAcc.: 0.96228; Mean BAcc: 0.88119; Mean sen.: 0.79477; Mean spe.: 0.91842; Avg. time: 2.36320
CrCb					Best BAcc.: 0.97951; Mean BAcc: 0.81375; Mean sen.: 0.85651; Mean spe.: 0.80762; Avg. time: 1.94140
					Best BAcc.: 0.94228; Mean BAcc: 0.78670; Mean sen.: 0.41777; Mean spe.: 0.86373; Avg. time: 2.6595
					Best BAcc.: 0.91606; Mean BAcc: 0.78668; Mean sen.: 0.41516; Mean spe.: 0.86913; Avg. time: 2.21520
Orig					Best BAcc.: 0.94253; Mean BAcc: 0.78293; Mean sen.: 0.27400; Mean spe.: 0.93489; Avg. time: 2.0863