Laser & Optoelectronics Progress, Volume. 59, Issue 12, 1210008(2022)
PCB Image-Denoising Algorithm Based on Image Difference and Residual Learning
Figures & Tables(11)
Fig. 6. Experimental results of network structure ablation. (a) SGD optimizer; (b) Adam optimizer
Fig. 7. Denoising comparison experiment results. (a) Original image; (b) Gaussian noise image; (c) median filter algorithm; (d) wavelet transform algorithm; (e) DnCNN; (f) proposed algorithm
Fig. 8. Gray value histograms of image denoising. (a) Original image; (b) Gaussian noise image; (c) median filter algorithm; (d) wavelet transform algorithm; (e) DnCNN; (f) proposed algorithm
Table 1. Number of images in noise dataset
View tableTable 1. Number of images in noise dataset
Dataset σ=15 dB σ=25 dB σ=35 dB Training data 7062 7062 7062 Testing data 334 334 334
Table 2. Denoising experiment results of different algorithms
View tableTable 2. Denoising experiment results of different algorithms
Noise level /dB Median filter algorithm Wavelet transform algorithm DnCNN Proposed algorithm PSNR SSIM PSNR SSIM PSNR SSIM PSNR SSIM σ=15 28.0318 0.7854 29.4439 0.8364 29.9231 0.8319 31.9231 0.8595 σ=25 26.6213 0.7291 27.1390 0.8253 28.5236 0.8023 31.6195 0.8457 σ=35 23.9974 0.6227 25.2371 0.7753 26.9512 0.7513 30.1610 0.8367
Table 3. Average time of PCB image denoising
View tableTable 3. Average time of PCB image denoising
Parameter Median filter algorithm Wavelet transform algorithm DnCNN Proposed algorithm Time /s 8.23 7.64 2.51 1.23
Tools
Get Citation
Copy Citation Text
Guangzai Ran, Lei Xu, Dashuang Li, Zhanling Guo. PCB Image-Denoising Algorithm Based on Image Difference and Residual Learning[J]. Laser & Optoelectronics Progress, 2022, 59(12): 1210008
Paper Information
Category: Image Processing
Received: Mar. 26, 2021
Accepted: Jun. 11, 2021
Published Online: May. 23, 2022
The Author Email: Lei Xu (xulei@scu.edu.cn)
© Copyright 2018-2021 | Chinese Laser Press.
All Rights Reserved 沪ICP备15018463号-20