Removal of Hyperspectral Stripe Noise Using Low-Pass Filtered Residual Images

Huihui Ju; Zhigang Liu; Jiangjun Jiang; Yang Wang

doi:10.3788/AOS201838.1228002

Journals >Acta Optica Sinica >Volume 38 >Issue 12 >Page 1228002 > Article

Acta Optica Sinica
Vol. 38, Issue 12, 1228002 (2018)

Removal of Hyperspectral Stripe Noise Using Low-Pass Filtered Residual Images

Huihui Ju^*, Zhigang Liu^*, Jiangjun Jiang, and Yang Wang

Author Affiliations

Institute of Nuclear Engineering, Rocket Force Engineering University, Xi'an, Shaanxi 710025, China

show less

DOI: 10.3788/AOS201838.1228002 Cite this Article Set citation alerts

Huihui Ju, Zhigang Liu, Jiangjun Jiang, Yang Wang. Removal of Hyperspectral Stripe Noise Using Low-Pass Filtered Residual Images[J]. Acta Optica Sinica, 2018, 38(12): 1228002 Copy Citation Text

show less

Fig. 1. Experimental results after 1000 Gaussian filtering. (a) Original strip noise image Y; (b) filtered image X; (c) residual image R

Download full size

Fig. 2. Separation results of strip noise and detail. (a) Residual image R; (b) details D; (c) strip noise S

Download full size

Fig. 3. Lenna images with strip noise of different degrees. (a) 0.01; (b) 0.02; (c) 0.05; (d) 0.1; (e) 0.2; (f) 0.5

Download full size

Fig. 4. Test images. (a) Test image-1; (b) test image-2

Download full size

Fig. 5. Stripe noise removing results of different algorithms on test image-1. (a) WFAF; (b) MDBC; (c) IDP; (d) RSLFRI

Download full size

Fig. 6. Stripe noise removing results of different algorithms on test image-2. (a) WFAF; (b) MDBC; (c) IDP; (d) RSLFRI

Download full size

Fig. 7. Comparison of mean column profiles before and after de-noising of test image-1

Download full size

Fig. 8. Comparison of mean column profiles before and after de-noising of test image-2

Download full size

Input: Y, K
Output: X
1. Initialization: X=Y, β=E^n×¹, ε=0.0001
2. While $\begin{matrix} {(β)}_{\infty} \end{matrix}$ >ε
3. R=X-K*X
4. β=R^TE^m×¹/m
5. F=X-E^m×¹β^T
6. X=F-μ_FE^m×n+μ_YE^m×n
7. End

Table 1. Steps of RSLFRI algorithm

Image	Parameter	σ
Image	Parameter	0.25	0.27	0.3	0.32	0.33	0.35	0.4	0.5	1	10
Lenna-1	I_RS	8×10^-7	8×10^-6	0.5639	0.6930	2.3333	4.5913	14.3913	26.3392	41.7984	44.7957
Lenna-1	I_IM	0.0003	0.0003	0.0002	0.0002	0.0006	0.0014	0.0050	0.0094	0.0151	0.0162
Lenna-2	I_RS	7×10^-7	0.0387	0.6214	0.9677	1.2007	1.8425	4.3673	7.3466	10.9396	11.6647
Lenna-2	I_IM	0.0014	0.0010	0.0005	0.0006	0.0008	0.0016	0.0051	0.0093	0.0144	0.0154
Lenna-3	I_RS	0.0245	0.3620	0.7204	0.8909	0.9608	1.0843	1.5794	2.1763	2.9538	3.0911
Lenna-3	I_IM	0.0065	0.0027	0.0015	0.0013	0.0014	0.0018	0.0048	0.0094	0.0157	0.0168
Lenna-4	I_RS	0.2235	0.4811	0.7024	0.8261	0.8646	0.9551	1.1540	1.3170	1.5199	1.5562
Lenna-4	I_IM	0.0162	0.0101	0.0057	0.0037	0.0033	0.0030	0.0057	0.0101	0.0162	0.0173
Lenna-5	I_RS	0.4640	0.6704	0.8168	0.8679	0.8830	0.9232	0.9835	1.0690	1.1410	1.1522
Lenna-5	I_IM	0.0373	0.0242	0.0157	0.0131	0.0123	0.0110	0.0116	0.0186	0.0282	0.0300
Lenna-6	I_RS	0.6227	0.7843	0.8780	0.9096	0.9219	0.9427	0.9904	1.0411	1.0974	1.1099
Lenna-6	I_IM	0.1394	0.0932	0.0693	0.0638	0.0623	0.0616	0.0751	0.1086	0.1454	0.1521

Table 2. Influences of filter standard deviation σ on the performance of RSLFRI algorithm

Image	Parameter	Initial image	WFAF	MDBC	IDP	RSLFRI
Test image-1	μ	0.4512	0.4512	0.4512	0.4512	0.4512
	δ	0.0133	0.0111	0.0101	0.0098	0.0112
	f_PSNR	20.9514	22.0721	21.8917	22.0234	22.1907
	H	6.8857	6.7179	6.6264	6.6159	6.7231
	$\begin{matrix} \bar{C} \end{matrix}$	1.0000	0.9827	0.9706	0.9777	0.9831
	$\begin{matrix} \bar{D} \end{matrix}$	0	0.0050	0.0054	0.0060	0.0048
Test image-2	μ	0.2176	0.2176	0.2176	0.2176	0.2176
	δ	0.0090	0.0090	0.0083	0.0081	0.0090
	f_PSNR	19.5468	19.5482	19.2601	19.3885	19.5560
	H	5.9748	5.9577	6.1546	5.8446	5.9477
	$\begin{matrix} \bar{C} \end{matrix}$	1.0000	0.9956	0.9906	0.9772	0.9970
	$\begin{matrix} \bar{D} \end{matrix}$	0	0.0002	0.0008	0.0006	0.0001

Table 3. Comparison of ability to maintain information for different algorithms

Huihui Ju, Zhigang Liu, Jiangjun Jiang, Yang Wang. Removal of Hyperspectral Stripe Noise Using Low-Pass Filtered Residual Images[J]. Acta Optica Sinica, 2018, 38(12): 1228002

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information