Image Denoising Using a Novel Deep Generative Network with Multiple Target Images and Adaptive Termination Condition

被引:5
作者
Chen, Shiming [1 ]
Xu, Shaoping [1 ]
Chen, Xiaoguo [1 ]
Li, Fen [1 ]
机构
[1] Nanchang Univ, Sch Informat Engn, Nanchang 330031, Peoples R China
来源
APPLIED SCIENCES-BASEL | 2021年 / 11卷 / 11期
关键词
image denoising; deep generative network; deep image prior; adaptive termination condition; multiple target images; denoising effect; QUALITY ASSESSMENT;
D O I
10.3390/app11114803
中图分类号
O6 [化学];
学科分类号
0703 ;
摘要
Image denoising, a classic ill-posed problem, aims to recover a latent image from a noisy measurement. Over the past few decades, a considerable number of denoising methods have been studied extensively. Among these methods, supervised deep convolutional networks have garnered increasing attention, and their superior performance is attributed to their capability to learn realistic image priors from a large amount of paired noisy and clean images. However, if the image to be denoised is significantly different from the training images, it could lead to inferior results, and the networks may even produce hallucinations by using inappropriate image priors to handle an unseen noisy image. Recently, deep image prior (DIP) was proposed, and it overcame this drawback to some extent. The structure of the DIP generator network is capable of capturing the low-level statistics of a natural image using an unsupervised method with no training images other than the image itself. Compared with a supervised denoising model, the unsupervised DIP is more flexible when processing image content that must be denoised. Nevertheless, the denoising performance of DIP is usually inferior to the current supervised learning-based methods using deep convolutional networks, and it is susceptible to the over-fitting problem. To solve these problems, we propose a novel deep generative network with multiple target images and an adaptive termination condition. Specifically, we utilized mainstream denoising methods to generate two clear target images to be used with the original noisy image, enabling better guidance during the convergence process and improving the convergence speed. Moreover, we adopted the noise level estimation (NLE) technique to set a more reasonable adaptive termination condition, which can effectively solve the problem of over-fitting. Extensive experiments demonstrated that, according to the denoising results, the proposed approach significantly outperforms the original DIP method in tests on different databases. Specifically, the average peak signal-to-noise ratio (PSNR) performance of our proposed method on four databases at different noise levels is increased by 1.90 to 4.86 dB compared to the original DIP method. Moreover, our method achieves superior performance against state-of-the-art methods in terms of popular metrics, which include the structural similarity index (SSIM) and feature similarity index measurement (FSIM). Thus, the proposed method lays a good foundation for subsequent image processing tasks, such as target detection and super-resolution.
引用
收藏
页数:21
相关论文
共 33 条
[1]   K-SVD: An algorithm for designing overcomplete dictionaries for sparse representation [J].
Aharon, Michal ;
Elad, Michael ;
Bruckstein, Alfred .
IEEE TRANSACTIONS ON SIGNAL PROCESSING, 2006, 54 (11) :4311-4322
[2]   Contour Detection and Hierarchical Image Segmentation [J].
Arbelaez, Pablo ;
Maire, Michael ;
Fowlkes, Charless ;
Malik, Jitendra .
IEEE TRANSACTIONS ON PATTERN ANALYSIS AND MACHINE INTELLIGENCE, 2011, 33 (05) :898-916
[3]   A non-local algorithm for image denoising [J].
Buades, A ;
Coll, B ;
Morel, JM .
2005 IEEE COMPUTER SOCIETY CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, VOL 2, PROCEEDINGS, 2005, :60-65
[4]   Regression-Based Neuro-Fuzzy Network Trained by ABC Algorithm for High-Density Impulse Noise Elimination [J].
Caliskan, Abdullah ;
Cil, Zeynel Abidin ;
Badem, Hasan ;
Karaboga, Dervis .
IEEE TRANSACTIONS ON FUZZY SYSTEMS, 2020, 28 (06) :1084-1095
[5]   Low-dose CT via convolutional neural network [J].
Chen, Hu ;
Zhang, Yi ;
Zhang, Weihua ;
Liao, Peixi ;
Li, Ke ;
Zhou, Jiliu ;
Wang, Ge .
BIOMEDICAL OPTICS EXPRESS, 2017, 8 (02) :679-694
[6]   Optimal Combination of Image Denoisers [J].
Choi, Joon Hee ;
Elgendy, Omar A. ;
Chan, Stanley H. .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2019, 28 (08) :4016-4031
[7]   Image denoising by sparse 3-D transform-domain collaborative filtering [J].
Dabov, Kostadin ;
Foi, Alessandro ;
Katkovnik, Vladimir ;
Egiazarian, Karen .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2007, 16 (08) :2080-2095
[8]   Nonlocally Centralized Sparse Representation for Image Restoration [J].
Dong, Weisheng ;
Zhang, Lei ;
Shi, Guangming ;
Li, Xin .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2013, 22 (04) :1618-1628
[9]   Image Quality Assessment Based on Multiscale Geometric Analysis [J].
Gao, Xinbo ;
Lu, Wen ;
Tao, Dacheng ;
Li, Xuelong .
IEEE TRANSACTIONS ON IMAGE PROCESSING, 2009, 18 (07) :1409-1423
[10]   Iterative PET Image Reconstruction Using Convolutional Neural Network Representation [J].
Gong, Kuang ;
Guan, Jiahui ;
Kim, Kyungsang ;
Zhang, Xuezhu ;
Yang, Jaewon ;
Seo, Youngho ;
El Fakhri, Georges ;
Qi, Jinyi ;
Li, Quanzheng .
IEEE TRANSACTIONS ON MEDICAL IMAGING, 2019, 38 (03) :675-685