Research Article

Color Face Recognition Based on Steerable Pyramid Transform and Extreme Learning Machines

Table 2

Performance comparison in different hybrid color spaces of LCSPT-ELM face recognition algorithm: the AR database.

ā€‰ex1ex2
ā€‰HS 0 HS 0

97.7298.4197.0296.3397.0296.7897.0896.2894.4097.08
95.2596.6194.7295.6494.6993.5194.5897.5895.5993.98
95.6398.4197.7194.9498.4195.8898.2697.0794.3997.07
96.3397.0297.7295.6496.3395.8897.0797.6795.8896.78
94.2495.6396.3390.7796.3393.5094.9994.3990.2394.99
94.0299.0197.0297.4198.4196.7898.1697.3795.9898.08
[5]93.7596.8297.7195.6495.6994.9997.8597.3794.9997.07
[5]93.5597.0297.7295.6497.7294.9995.5995.5992.9195.29
[10]94.2597.0297.7295.6497.7294.6998.5697.3794.1097.97
[8]94.9497.7197.7196.3398.4194.3996.1895.8893.8095.88
[9]96.3397.7297.7295.6498.4194.6996.2896.1893.8096.18
[5]97.0297.7297.0294.9497.0296.1896.4897.3794.1098.26
96.3399.1199.8097.7299.1196.7898.8696.2696.7899.45
95.6499.1199.8097.0299.1197.0799.1698.5696.7899.16
95.6499.1199.8098.4199.1196.1899.1698.5696.7899.16

Bold values specify the hybrid color spaces and their subbands with the highest correctness.