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

Table 1

Performance comparison in color component images of LCSPT-ELM face recognition algorithm: the AR database.

	ex1					ex2
	HS	−𝜋/4	0	𝜋/4	𝜋/2	HS	−𝜋/4	0	𝜋/4	𝜋/2
𝑅	90.97	95.13	94.44	87.50	94.44	91.96	91.07	91.36	86.30	91.36
𝐺	91.66	94.44	93.05	90.97	93.05	90.77	91.66	91.66	85.71	91.07
𝐵	91.66	91.66	93.13	92.64	94.11	93.75	93.15	93.75	91.07	94.04
𝑌	93.75	94.44	93.05	91.66	94.44	94.04	93.75	94.64	90.47	94.04
𝐶𝑏	86.11	90.97	90.97	86.11	91.67	85.71	89.28	90.17	87.20	92.26
𝐶𝑟	86.80	92.36	93.75	88.19	93.05	86.90	92.55	91.66	88.39	92.26
𝑆	89.58	91.66	95.83	90.97	94.44	90.17	94.34	94.94	89.58	94.94
𝐻	85.41	82.63	84.02	75.00	90.27	87.20	85.41	86.30	77.08	86.90
𝑉	90.97	93.05	90.97	87.50	90.97	88.09	86.60	86.90	83.63	88.09
𝑌	90.97	93.05	93.75	87.50	92.36	91.36	92.26	92.85	88.69	92.26
𝐼	88.89	90.27	95.13	90.27	91.66	87.20	94.04	93.75	88.09	91.36
𝑄	74.30	77.77	82.63	74.30	81.25	71.42	74.10	80.95	74.70	76.78
𝐿^∗	91.66	95.13	93.05	90.97	93.75	91.07	94.04	93.15	89.88	93.75
𝑎^∗	82.63	90.27	92.36	86.80	90.97	83.63	88.98	91.66	84.52	87.50
𝑏^∗	93.62	90.68	92.15	92.15	92.15	85.71	89.28	91.36	84.82	90.47
𝐺^𝑛	90.97	92.36	93.75	88.88	94.44	91.36	92.55	92.26	88.39	92.26
𝐵^𝑛	93.05	92.36	96.52	91.66	93.75	92.85	93.45	94.64	90.47	93.15
𝑋^𝑛	92.36	94.44	94.44	90.27	94.44	94.04	93.75	94.04	90.77	93.75
𝑌^𝑛	87.50	89.58	90.97	86.80	88.19	89.28	88.98	91.66	86.90	89.88
𝑍^𝑛	90.97	92.36	96.52	92.36	94.44	93.45	93.15	94.04	90.47	94.04
𝑋	89.58	94.44	93.75	86.80	90.97	90.47	92.55	92.26	88.69	92.85
𝑍	92.36	93.05	95.13	88.19	93.75	92.85	92.55	92.26	88.09	93.45
𝐺𝑟	85.41	92.36	94.44	87.50	92.36	82.73	93.75	90.17	86.30	89.28
𝐵𝑟	89.58	92.36	92.36	88.88	94.44	87.79	91.36	93.45	87.79	94.94

Bold values mean the color components with correctness over 90% for all subbands and the subbands with the highest correctness.