Research Article

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

Table 6

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

fb setdup1 set
HSπ/40π/4π/2HSπ/40π/4π/2

97.7997.79097.3096.8196.8176.8673.9274.4173.4375.88
90.4296.85097.7193.3495.9280.3280.3480.2478.1185.17
99.2697.3098.7798.2898.7782.7484.7084.7081.2784.21
96.3297.3098.7796.3297.7975.8877.3582.2576.8680.29
91.4294.8596.8193.3896.3270.0072.9476.8675.8878.33
92.1297.8897.3097.1798.8180.7483.2383.4182.4384.84
[5]93.8994.8597.8393.8795.3280.6884.7083.2380.8079.25
[5]92.8994.8596.8193.8796.3274.9073.4375.8875.8879.31
[10]99.7096.8197.3095.8397.7980.7884.7084.2179.8082.25
[8]95.8396.8198.2896.8198.2875.3980.2982.7479.3179.80
[9]98.2897.3097.7997.7996.8176.3779.8081.7679.3181.76
[5]94.3694.8595.8393.8796.3275.8875.8877.3574.9077.84
96.8198.2898.7798.2899.7585.1984.7084.7083.2389.11
97.7998.2898.7799.2699.7582.7484.2183.7281.2789.11
97.7998.2898.7799.2699.7583.7283.7285.1982.7489.11

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