Skip to main content
. Author manuscript; available in PMC: 2012 Jun 1.
Published in final edited form as: Neurocomputing (Amst). 2011 Jun 1;74(12-13):2184–2192. doi: 10.1016/j.neucom.2011.02.014

Table 2.

Recognition rates of GF, SURF and MRGF methods. Numbers of histogram bins n = 8, 10 and 16 were tested. Columns from left to right: Sitting-up(SU), sitting-still(SS), walking(WK), bowing(BW), crouching(CR), waist exercise(WE) and total average(ALL).

SU SS WK BW CR WE ALL
GF n = 8 91% 72% 75% 70% 86% 91% 81%
n = 10 86% 93% 84% 82% 94% 91% 88%
n = 16 87% 94% 90% 86% 98% 96% 92%
SURF n = 8 95% 94% 82% 88% 84% 94% 90%
n = 10 96% 96% 90% 89% 93% 95% 93%
n = 16 98% 98% 95% 95% 97% 97% 97%
MRGF n = 8 93% 88% 80% 82% 91% 94% 88%
n = 10 93% 98% 90% 88% 97% 92% 93%
n = 16 95% 98% 92% 92% 98% 97% 95%