Method of joint wavelet thresholding and F-NLM de-noising for high-resolution SAR ship detection

Table 1. Test environment configuration parameters
View table
View in Article
Table 1. Test environment configuration parameters
硬件配置或软件环境参数或版本
CPU i7-12700
RAM 16 GB
GPU NVDIA RTX3050
操作系统 Windows 10
编程语言及软件 Python3.8，Matlab2019b
架构 CUDA10.1，cuDNN7.6

Table 2. SSDD dataset information
View table
View in Article
Table 2. SSDD dataset information
数据集传感器分辨率/m 极化方式数量
SSDD RadarSat-2 1～15 HH，HVVV，VH
TerraSAR-X 1160
Sentinel-1

Table 3. Comparison of de-noising performance in terms of P_SNR and average run time using different methods

View table

View in Article

Table 3. Comparison of de-noising performance in terms of P_SNR and average run time using different methods

输入图像	噪声方差	不同方法获得的峰值信噪比R_PSNR/dB
输入图像	噪声方差	均值滤波	Lee滤波	小波阈值	NLM滤波	融合去噪
A₁	20	19.26	20.19	21.12	23.48	23.71
A₁	40	16.96	17.30	17.91	19.15	20.94
A₂	20	21.29	22.24	23.37	25.29	25.53
A₂	40	18.55	19.51	19.98	20.32	21.83
A₃	20	21.78	22.71	23.33	25.28	25.52
A₃	40	18.89	19.83	20.04	20.43	22.03
B₁	20	17.43	18.50	19.00	20.63	20.91
B₁	40	15.56	15.85	16.58	16.65	18.39
B₂	20	17.80	18.81	19.03	20.68	21.43
B₂	40	15.57	16.38	16.54	17.54	19.65
B₃	20	16.17	17.80	18.95	20.15	20.62
B₃	40	15.12	16.13	16.98	17.57	20.33
C₁	20	15.74	16.79	18.43	19.36	19.60
C₁	40	14.26	15.29	15.86	16.64	18.39
C₂	20	17.14	18.15	19.03	20.74	21.00
C₂	40	15.32	16.31	16.38	17.56	19.66
C₃	20	18.53	19.54	19.78	21.34	22.42
C₃	40	16.38	17.40	16.97	18.55	20.32
R_PSNR平均值/dB		17.32	18.27	18.85	20.08	21.24
平均运行时间/s		0.625	0.609	1.068	56.375	2.609

Table 4. Detection performance comparison of different datasets using the same detection model
View table
View in Article
Table 4. Detection performance comparison of different datasets using the same detection model
数据集 P/% R/% AP₅₀/% F1分数 r_FAR/%
SSDD 94.57 96.14 96.44 0.9715 6.20
D-SSDD 95.59 96.61 97.63 0.9798 4.15

Table 5. Comparison of target detection performance using different algorithms

View table

View in Article

Table 5. Comparison of target detection performance using different algorithms

方法	P/%	R/%	AP₅₀/%	F1分数	r_FAR/%	模型体积/MB
Faster R-CNN	89.33	90.12	93.25	0.930 4	9.75	156.43
EffcientDet	94.16	95.81	96.07	0.965 1	7.03	89.71
YOLOv5	94.57	96.14	96.44	0.971 5	6.20	14.40
YOLOv7	95.37	96.47	97.36	0.977 6	4.15	27.00
YOLOv7+Bi-FPN	96.62	96.85	98.69	0.98 79	2.37	27.30

Tools

Get Citation

Copy Citation Text

Liang TONG, Dan LIU, Zhongbo PENG, Han ZOU, Lumeng WANG, Chunyu ZHANG. Method of joint wavelet thresholding and F-NLM de-noising for high-resolution SAR ship detection[J]. Chinese Journal of Ship Research, 2024, 19(6): 275

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites

Paper Information

Category: Weapon, Electronic and Information System

Received: Jul. 26, 2023

Accepted: --

Published Online: Mar. 14, 2025

The Author Email:

DOI:10.19693/j.issn.1673-3185.03477

Topics

laser devices and laser physics

Lasers and Laser Optics

Laser physics

laser manufacturing

Instrumentation, Measurement and Metrology

Table 1. Test environment configuration parameters

Table 1. Test environment configuration parameters

Table 2. SSDD dataset information

Table 2. SSDD dataset information

Table 3. Comparison of de-noising performance in terms of PSNR and average run time using different methods

Table 3. Comparison of de-noising performance in terms of PSNR and average run time using different methods

Table 4. Detection performance comparison of different datasets using the same detection model

Table 4. Detection performance comparison of different datasets using the same detection model

Table 5. Comparison of target detection performance using different algorithms

Table 5. Comparison of target detection performance using different algorithms

微信扫一扫：分享

Table 3. Comparison of de-noising performance in terms of P_SNR and average run time using different methods

Table 3. Comparison of de-noising performance in terms of P_SNR and average run time using different methods