Spatio-spectral binary patterns based on multispectral filter arrays for texture classification

Sofiane Mihoubi; Olivier Losson; Benjamin Mathon; Ludovic Macaire

doi:10.1364/JOSAA.35.001532

Journal of the Optical Society of America A
Vol. 35,
Issue 9,
pp. 1532-1542
(2018)
•https://doi.org/10.1364/JOSAA.35.001532

Spatio-spectral binary patterns based on multispectral filter arrays for texture classification

Sofiane Mihoubi, Olivier Losson, Benjamin Mathon, and Ludovic Macaire

Not Accessible

Your library or personal account may give you access

Get PDF
Email
Share
Get Citation
Copy Citation Text
Sofiane Mihoubi, Olivier Losson, Benjamin Mathon, and Ludovic Macaire, "Spatio-spectral binary patterns based on multispectral filter arrays for texture classification," J. Opt. Soc. Am. A 35, 1532-1542 (2018)

Export Citation
- BibTex
- Endnote (RIS)
- HTML
- Plain Text
Citation alert
Save article

Check for updates

Abstract

To discriminate gray-level texture images, spatial texture descriptors can be extracted using the local binary pattern (LBP) operator. This operator has been extended to color images at the expense of increased memory and computation requirements. Some authors propose to compute texture descriptors directly from raw images provided through a Bayer color filter array, which both avoids the demosaicking step and reduces the descriptor size. Recently, multispectral snapshot cameras have emerged to sample more than three wavelength bands using a multispectral filter array. Such cameras provide a raw image in which a single spectral channel value is available at each pixel. In this paper we design a local binary pattern operator that jointly extracts the spatial and spectral texture information directly from a raw image. Extensive experiments on a large dataset show that the proposed descriptor has both reduced computation cost and high discriminative power with regard to classical LBP descriptors applied to demosaicked images.

Full Article | PDF Article

More Like This

Combining local binary patterns and local color contrast for texture classification under varying illumination

Claudio Cusano, Paolo Napoletano, and Raimondo Schettini
J. Opt. Soc. Am. A 31(7) 1453-1461 (2014)

Snapshot multispectral polarization imaging using a photonic crystal filter array

Kazuma Shinoda, Yasuo Ohtera, and Madoka Hasegawa
Opt. Express 26(12) 15948-15961 (2018)

Structure descriptor based on just noticeable difference for texture image classification

Xikui Miao, Wei Zhao, Xiaolong Li, and Xiaoyu Yang
Appl. Opt. 58(24) 6504-6512 (2019)

Previous Article Next Article

Cited By

You do not have subscription access to this journal. Cited by links are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Figures (8)

You do not have subscription access to this journal. Figure files are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Tables (3)

You do not have subscription access to this journal. Article tables are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

Equations (18)

You do not have subscription access to this journal. Equations are available to subscribers only. You may subscribe either as an Optica member, or as an authorized user of your institution.

Contact your librarian or system administrator
or
Login to access Optica Member Subscription

MSFA	$d = 1$	$d = 2$	$d = 3$
Vis5	3 or 5	1 or 2	3 or 5
VisNIR8	6	2	6
IMEC16	8	3	8
IMEC25	8	8	8

		Feature Size		Number of Operations
Approach	Eq.	Hist. Size	Number of Hist.	Demosaicking	Feature Computation
Marginal LBPs [17]	Eq. (1)	$2^{P}$	$K$	$4 K - 4$	$24 K$
Moment LBPs [19]	Eq. (3), Eq. (4)	$2^{P}$	3	$4 K - 4$	$K^{2} + 10 K + 69$
maLBP [20]	Eq. (5)	$2^{P}$	$K + 1$	$4 K - 4$	$17 K + 8$
mdLBP [20]	Eq. (6)	$2^{P}$	$2^{K}$	$4 K - 4$	$2^{K} + 24 K + 7$
Cusano LBPs [21]	Eq. (8)	$2^{P}$	2	$4 K - 4$	$15 K + 27$
Lee LBPs [22]	Eq. (9)	$2^{P}$	$1 + K (K - 1)$	$4 K - 4$	$27 K^{2} - 26 K + 24$
OBLBP [23]	Eq. (10)	$2^{P}$	$K^{2}$	$4 K - 4$	$24 K^{2}$
IOBLBP [24]	Eq. (11)	$2^{P + 1}$	$K^{2}$	$4 K - 4$	$27 K^{2} + 10 K$
MLBP (ours)	Eq. (14), Eq. (15)	$2^{P}$	$K$	0	24

MSFA: Vis5 ( $K = 5$ )
Approach	E						D65						A
	$d = 1$		$d = 2$		$d = 3$		$d = 1$		$d = 2$		$d = 3$		$d = 1$		$d = 2$		$d = 3$
	WB	PPID	WB	PPID	WB	PPID	WB	PPID	WB	PPID	WB	PPID	WB	PPID	WB	PPID	WB	PPID
Marginal LBPs [17]	87.71	86.95	88.39	87.50	88.67	88.46	87.36	87.23	88.60	88.26	88.46	88.60	84.89	84.62	87.71	87.84	88.05	88.19
Moment LBPs [19]	90.11	88.05	88.26	88.67	86.74	86.13	90.32	88.53	89.29	89.01	88.53	87.36	87.98	87.16	86.74	86.74	85.99	85.71
maLBP [20]	82.83	84.55	86.06	86.40	85.65	86.95	81.11	82.01	84.34	84.89	85.30	86.13	81.66	82.21	85.99	86.20	86.88	88.32
Cusano LBPs [21]	87.64	87.36	88.60	88.26	88.80	89.15	86.88	87.36	88.46	88.39	89.29	89.22	88.80	89.01	90.52	90.25	90.32	90.87
Lee LBPs [22]	93.41	93.75	95.40	95.12	95.60	94.85	92.99	93.13	94.57	94.44	95.47	95.47	91.48	91.96	94.23	93.82	94.64	94.37
OBLBP [23]	97.39	97.46	97.39	97.53	97.12	97.46	96.84	96.70	97.18	97.39	97.12	97.39	94.64	94.51	95.67	95.95	96.29	96.57
IOBLBP [24]	96.02	95.95	95.60	96.29	96.70	96.09	95.05	94.57	95.81	96.15	96.36	96.15	91.55	92.24	93.06	93.41	94.57	94.09
MLBP (Ours)	93.82		92.86		94.99		94.92		92.45		96.63		84.48		92.99		89.90
MSFA: VisNIR8 ( $K = 8$ )
Marginal LBPs [17]	84.89	86.40	85.44	85.78	86.06	86.13	87.09	88.94	87.50	88.67	87.64	88.53	84.00	85.51	85.03	85.37	85.78	85.51
Moment LBPs [19]	86.26	86.40	85.85	85.37	86.06	84.75	84.89	84.62	85.71	84.55	85.51	85.23	82.76	83.04	83.93	83.79	84.20	83.79
maLBP [20]	77.75	77.88	79.46	79.33	82.55	79.95	77.54	79.33	80.22	81.11	79.95	79.95	78.23	76.99	79.12	78.78	81.46	80.29
Cusano LBPs [21]	82.49	82.55	83.93	83.72	84.34	85.03	85.58	85.51	87.16	87.09	87.16	87.57	80.77	80.43	82.42	82.01	82.49	83.17
Lee LBPs [22]	91.28	91.69	93.96	93.89	95.05	95.26	89.35	91.41	91.83	91.69	93.48	93.41	91.62	92.86	93.82	93.89	95.26	94.16
OBLBP [23]	96.29	96.91	96.63	96.57	96.57	96.63	96.09	96.22	96.15	95.67	96.09	95.54	95.19	95.40	95.47	95.12	95.12	95.33
IOBLBP [24]	94.09	93.61	94.64	94.30	94.44	94.57	94.30	94.02	93.82	93.61	94.23	93.48	92.10	93.34	92.65	92.31	92.86	92.51
MLBP (Ours)	93.75		95.67		94.57		94.57		94.71		94.64		93.06		94.02		93.20
MSFA: IMEC16 ( $K = 16$ )
Marginal LBPs [17]	81.59	85.10	83.79	86.54	83.86	86.20	81.11	84.89	83.72	86.95	85.30	87.16	80.70	84.89	83.10	85.92	85.71	87.50
Moment LBPs [19]	85.30	89.56	87.50	90.32	87.71	90.04	84.41	88.67	88.12	89.42	87.02	89.90	84.34	88.05	88.05	89.84	88.05	89.15
maLBP [20]	74.93	78.57	75.21	76.17	76.51	78.57	73.56	79.12	75.41	77.13	75.34	80.08	75.07	77.13	74.52	73.97	76.51	74.86
Cusano LBPs [21]	84.07	86.40	86.20	87.98	85.92	87.98	85.30	86.20	86.33	87.91	86.54	87.02	85.44	86.61	87.02	88.19	87.16	88.67
Lee LBPs [22]	89.77	95.40	92.45	96.02	93.68	95.95	89.22	95.26	92.58	95.95	93.41	95.81	87.09	94.44	90.93	94.09	92.93	94.64
OBLBP [23]	96.91	96.57	96.22	96.77	96.09	96.50	97.25	97.60	97.05	97.05	96.84	96.98	95.67	96.09	95.47	95.81	95.05	95.95
IOBLBP [24]	95.40	96.36	95.19	95.88	95.33	95.47	96.02	96.91	95.88	96.57	95.74	96.29	93.54	94.71	94.44	94.78	93.61	94.64
MLBP (Ours)	96.22		90.87		96.29		97.32		94.92		97.05		95.33		90.32		95.19
MSFA: IMEC25 ( $K = 25$ )
Marginal LBPs [17]	73.15	78.57	74.93	78.71	75.62	78.50	73.63	78.71	75.76	80.01	77.47	79.81	72.87	78.09	74.45	78.43	75.82	77.54
Moment LBPs [19]	79.46	83.17	80.56	83.72	80.77	84.27	78.98	84.20	81.25	85.51	81.46	85.16	78.71	83.31	80.15	83.93	80.36	84.07
maLBP [20]	63.67	78.64	61.26	78.50	67.03	76.79	68.27	78.09	62.77	77.06	65.73	75.14	64.08	77.88	60.37	78.37	67.17	75.55
Cusano LBPs [21]	77.13	81.80	78.57	82.69	77.68	81.39	75.34	80.15	76.51	82.35	78.98	82.14	76.65	81.18	78.64	81.94	77.75	80.70
Lee LBPs [22]	87.77	92.58	92.03	93.89	92.79	94.57	89.49	94.16	91.14	94.78	92.31	94.92	88.12	91.96	91.96	93.54	92.65	94.23
OBLBP [23]	94.02	95.05	93.41	94.92	93.48	94.71	94.99	95.88	94.51	95.95	94.51	95.60	93.75	94.51	93.13	94.71	93.06	94.16
IOBLBP [24]	92.93	94.37	91.83	93.68	91.21	92.58	94.09	94.92	93.82	95.33	92.79	94.85	92.51	93.48	91.35	93.54	90.73	92.24
MLBP (Ours)	94.30		93.13		92.03		95.40		95.33		94.78		94.71		91.76		92.24

MSFA	$d = 1$	$d = 2$	$d = 3$
Vis5	3 or 5	1 or 2	3 or 5
VisNIR8	6	2	6
IMEC16	8	3	8
IMEC25	8	8	8

		Feature Size		Number of Operations
Approach	Eq.	Hist. Size	Number of Hist.	Demosaicking	Feature Computation
Marginal LBPs [17]	Eq. (1)	$2^{P}$	$K$	$4 K - 4$	$24 K$
Moment LBPs [19]	Eq. (3), Eq. (4)	$2^{P}$	3	$4 K - 4$	$K^{2} + 10 K + 69$
maLBP [20]	Eq. (5)	$2^{P}$	$K + 1$	$4 K - 4$	$17 K + 8$
mdLBP [20]	Eq. (6)	$2^{P}$	$2^{K}$	$4 K - 4$	$2^{K} + 24 K + 7$
Cusano LBPs [21]	Eq. (8)	$2^{P}$	2	$4 K - 4$	$15 K + 27$
Lee LBPs [22]	Eq. (9)	$2^{P}$	$1 + K (K - 1)$	$4 K - 4$	$27 K^{2} - 26 K + 24$
OBLBP [23]	Eq. (10)	$2^{P}$	$K^{2}$	$4 K - 4$	$24 K^{2}$
IOBLBP [24]	Eq. (11)	$2^{P + 1}$	$K^{2}$	$4 K - 4$	$27 K^{2} + 10 K$
MLBP (ours)	Eq. (14), Eq. (15)	$2^{P}$	$K$	0	24

Abstract

Cited By

Figures (8)

Tables (3)

Equations (18)

Journal of the Optical Society of America A