Multibody Nonrigid Structure from Motion Segmentation Based on Sparse Subspace Clustering

Table 2

Accuracy () of five algorithms in the 15 datasets synthesised.

Dataset		SSC	EnSC	k-SSC	SSSC	MNR_S
Three objects	Shark+dance+yoga (207)	0.992	1.000	0.903	0.928	0.993
	Yoga+stretch+pickup (123)	0.919	0.959	0.920	0.903	0.986
	Dance+pickup+stretch (157)	0.828	0.942	0.940	0.890	0.987
	Shark+walking+drink (187)	1.000	0.983	0.968	0.902	0.989
	Dance+yoga+drink (157)	0.917	0.920	0.902	0.886	0.980
Four objects	Shark+dance+yoga+walking (262)	1.000	0.996	0.901	0.907	0.980
	Dance+stretch+pickup+shark (248)	0.714	0.810	0.779	0.800	0.968
	Walking+face+pickup+drink (174)	0.701	0.850	——	0.813	0.948
	Dance+yoga+drink+walking (212)	0.755	0.780	——	0.834	0.984
	Dance+yoga+stretch+pickup (198)	0.858	0.803	0.737	0.785	0.989
Five objects	Shark+dance+yoga+pickup+walking(303)	0.736	0.799	0.742	0.783	0.939
	Shark+yoga+pickup+stretch+dance (289)	0.773	0.775	——	0.757	0.980
	Shark+yoga+walking+drink+dance (303)	0.731	0.791	0.751	0.698	0.970
	Dance+yoga+stretch+pickup+walking (253)	0.731	0.704	0.681	0.708	0.964
	Drink+walking+yoga+pickup+stretch(219)	0.739	0.794	——	0.750	0.937