Multiscale texture retrieval based on low-dimensional and rotation-invariant features of curvelet transform

EURASIP Journal on Image and Video Processing

Table 1 Classification rates (%)

	Classification	TC10	TC12-horizon	TC12-t184	KTH-TIPS	Brodatz	UIUCTex
Train 10	$F_{KDE}^{PO}$	91.94	92.06	90.24	80.70	95.39	65.63
	$F_{μσ}^{PO}$	91.15	91.48	88.37	75.41	92.73	65.21
	$F_{μσ}^{', PO}$	89.47	89.39	85.78	69.21	90.86	52.10
	F ^[13]	82.97	82.48	78.23	68.11	82.27	46.82
Train 20	$F_{KDE}^{PO}$	94.87	95.66	93.59	86.69	97.24	71.41
	$F_{μσ}^{PO}$	94.80	95.15	93.32	82.56	95.81	70.36
	$F_{μσ}^{', PO}$	92.83	92.81	90.22	76.36	94.23	57.95
	F ^[13]	88.72	87.94	84.78	76.20	88.69	51.73
Train 30	$F_{KDE}^{PO}$	96.60	96.64	95.25	88.31	98.02	74.08
	$F_{μσ}^{PO}$	96.12	96.50	95.06	87.92	96.88	73.79
	$F_{μσ}^{', PO}$	94.62	94.46	91.91	79.57	95.72	61.81
	F ^[13]	90.77	90.99	87.39	77.37	91.08	53.95
Train 40	$F_{KDE}^{PO}$	97.53	97.53	95.94	90.83	98.47
	$F_{μσ}^{PO}$	96.86	97.49	95.86	89.42	97.70
	$F_{μσ}^{', PO}$	95.24	95.41	92.85	83.27	96.71
	F ^[13]	92.58	92.36	88.89	82.15	93.04
Train 50	$F_{KDE}^{PO}$	97.82	98.05	96.62	92.19	98.91
	$F_{μσ}^{PO}$	97.60	97.82	96.46	91.10	98.05
	$F_{μσ}^{', PO}$	95.90	95.99	93.74	84.32	97.53
	F ^[13]	93.30	93.20	90.23	82.77	94.01
Train 60	$F_{KDE}^{PO}$	98.00	98.40	97.20	92.84	99.04
	$F_{μσ}^{PO}$	97.73	98.22	96.92	91.52	98.37
	$F_{μσ}^{', PO}$	96.16	96.40	94.54	84.00	97.64
	F ^[13]	94.10	93.74	91.03	83.14	94.76
Train 70	$F_{KDE}^{PO}$	98.33	98.46	97.44	93.09	99.18
	$F_{μσ}^{PO}$	98.22	98.29	97.28	92.36	98.72
	$F_{μσ}^{', PO}$	96.52	96.44	94.86	86.00	98.10
	F ^[13]	94.40	94.63	91.79	85.64	95.34

All the databases are evaluated for various (10,20,…,70) number of training images except the UIUCTex database since it only includes 40 images per class.