SAR Image Sparse Denoising Based on Blind Estimation and Bilateral Filtering

Yu Sun; Zhihui Xin; Penghui Huang; Zhixu Wang; Jiayu Xuan

doi:10.3788/LOP222462

Journals >Laser & Optoelectronics Progress >Volume 60 >Issue 16 >Page 1610001 > Article

Laser & Optoelectronics Progress
Vol. 60, Issue 16, 1610001 (2023)

SAR Image Sparse Denoising Based on Blind Estimation and Bilateral Filtering

Yu Sun¹, Zhihui Xin^1,*, Penghui Huang², Zhixu Wang¹, and Jiayu Xuan¹

Author Affiliations

¹Yunnan Key Laboratory of Opto-Electronic Information Technology, School of Physics and Electronic Information, Yunnan Normal University, Kunming 650500, Yunnan, China

²Department of Electronic Engineering, Shanghai Jiao Tong University, Shanghai 200240, China

show less

DOI: 10.3788/LOP222462 Cite this Article Set citation alerts

Yu Sun, Zhihui Xin, Penghui Huang, Zhixu Wang, Jiayu Xuan. SAR Image Sparse Denoising Based on Blind Estimation and Bilateral Filtering[J]. Laser & Optoelectronics Progress, 2023, 60(16): 1610001 Copy Citation Text

show less

Fig. 1. Flow chart of the SR-BBF

Download full size

Fig. 2. Experimental images. (a) Scene 1; (b) scene 2; (c) scene 3

Download full size

Fig. 3. Denoising results of scene 1. (a) Noise image; (b) Lee filtering; (c) Frost filtering; (d) BM3D filtering; (e) BF; (f) SR; (g) SR-Bayes; (h) proposed method

Download full size

Fig. 4. Denoising results of scene 2. (a) Noise image; (b) Lee filtering; (c) Frost filtering; (d) BM3D filtering; (e) BF; (f) SR; (g) SR-Bayes; (h) proposed method

Download full size

Fig. 5. Denoising results of scene 3. (a) Noise image; (b) Lee filtering; (c) Frost filtering; (d) BM3D filtering; (e) BF; (f) SR; (g) SR-Bayes; (h) proposed method

Download full size

Fig. 6. Noise-adding images under different

σ

. (a) Original image; (b)

σ = 0.5

; (c)

σ = 1

; (d)

σ = 1.5

Download full size

Fig. 7. Denoising results under

σ = 0.5

. (a) Noise image; (b) Lee filtering; (c) Frost filtering; (d) BM3D filtering; (e) BF; (f) SR; (g) SR-Bayes; (h) proposed method

Download full size

Fig. 8. Denoising results under

σ = 1

. (a) Noise image; (b) Lee filtering; (c) Frost filtering; (d) BM3D filtering; (e) BF; (f) SR; (g) SR-Bayes; (h) proposed method

Download full size

Fig. 9. Denoising results under

σ = 1.5

. (a) Noise image; (b) Lee filtering; (c) Frost filtering ; (d) BM3D filtering; (e) BF; (f) SR; (g) SR-Bayes; (h) proposed method

Download full size

Method	PSNR /dB	ENL	EPI	SSIM
Unfiltered	21.18	20.19
Lee	23.57	72.21	0.63	0.53
Frost	23.40	49.79	0.70	0.51
BM3D	23.13	120.03	0.77	0.56
BF	24.30	39.64	0.87	0.66
SR	24.53	32.26	0.86	0.69
SR-Bayes	24.49	43.86	0.79	0.61
Proposed method	24.94	131.02	0.80	0.61

Table 1. Comparison of denoising quality indicators of scene 1

Method	PSNR /dB	ENL	EPI	SSIM
Unfiltered	13.66	18.05
Lee	14.23	75.48	0.43	0.51
Frost	14.84	58.24	0.54	0.52
BM3D	14.68	76.26	0.51	0.64
BF	15.66	41.46	0.82	0.79
SR	14.51	44.87	0.74	0.70
SR-Bayes	15.89	81.20	0.78	0.68
Proposed method	16.38	93.28	0.80	0.75

Table 2. Comparison of denoising quality indicators of scene 2

Method	PSNR /dB	ENL	EPI	SSIM
Unfiltered	27.21	20.41
Lee	27.27	56.01	0.47	0.79
Frost	28.07	49.09	0.55	0.82
BM3D	27.82	75.64	0.50	0.81
BF	28.39	58.16	0.78	0.89
SR	27.55	70.02	0.72	0.85
SR-Bayes	27.46	72.43	0.75	0.82
Proposed method	28.72	87.62	0.81	0.83

Table 3. Comparison of denoising quality indicators of scene 3

Method	$σ = 0.5$				$σ = 1$				$σ = 1.5$
Method	PSNR /dB	ENL	EPI	SSIM	PSNR /dB	ENL	EPI	SSIM	PSNR /dB	ENL	EPI	SSIM
Unfiltered	24.12	1.85	0.54	0.69	21.57	0.92	0.37	0.44	20.13	0.73	0.30	0.32
Lee	24.39	5.59	0.68	0.74	23.15	3.68	0.76	0.77	21.05	3.91	0.64	0.69
Frost	24.83	4.32	0.65	0.81	22.58	2.58	0.69	0.67	21.37	2.41	0.56	0.55
BM3D	25.09	6.72	0.75	0.75	22.53	4.58	0.69	0.65	21.37	5.51	0.60	0.62
BF	25.25	8.19	0.73	0.74	22.68	1.75	0.47	0.62	20.52	1.09	0.36	0.44
SR	25.40	23.95	0.89	0.84	23.22	13.07	0.67	0.72	21.37	9.09	0.60	0.63
SR-Bayes	25.41	25.35	0.90	0.85	23.16	11.62	0.66	0.72	21.42	9.71	0.61	0.63
Proposed method	25.32	28.13	0.90	0.83	23.40	16.60	0.73	0.71	22.23	14.78	0.70	0.60

Table 4. Comparison of image quality indicators after filtering under different noise levels

Parameter	Scene 1	Scene 2	Scene 3	Noise-adding image1	Noise-adding image 2	Noise-adding image 3
BI	0.018	0.058	0.026	0.024	0.015	0.058

Table 5. Image blockiness index of the proposed method

Yu Sun, Zhihui Xin, Penghui Huang, Zhixu Wang, Jiayu Xuan. SAR Image Sparse Denoising Based on Blind Estimation and Bilateral Filtering[J]. Laser & Optoelectronics Progress, 2023, 60(16): 1610001

Download Citation

Set citation alerts for the article

Tools

Set citation alerts for the article

Save the article for my favorites

Paper Information

微信扫一扫：分享

微信扫一扫：分享