WTCRNet: a wavelet transform and contrastive regularization network for sonar denoising by self-supervision

Abstract

In the field of underwater acoustics, forward-looking sonar represents a pivotal tool for acquiring subaqueous imagery. However, this technique is susceptible to the inherent ambient noise prevalent in underwater environments, resulting in degraded image quality. A notable challenge in this domain is the scarcity of pristine image exemplars, making it difficult to apply many advanced deep denoising networks for the purification of sonar images. To address this issue, the study introduces a novel self-supervised methodology specifically designed for denoising forward-looking sonar images. The proposed model employs a blind-spot network architecture to reconstruct unblemished images. Additionally, it integrates wavelet transform technology within a convolutional neural network (CNN) framework, combining frequency and structural information. Furthermore, the model incorporates contrastive regularization to augment denoising efficiency. This innovative denoising network, which leverages wavelet transform and contrastive regularization (CR), is henceforth referred to as WTCRNet. To evaluate the performance of WTCRNet, this study constructs a dual dataset comprising both simulated and authentic forward-looking sonar images, thereby furnishing a comprehensive dataset for network training and evaluation. Empirical assessments conducted on these datasets demonstrate that WTCRNet substantially outperforms existing denoising methodologies by effectively mitigating noise. The code is available at https://gitee.com/sichengling/wtcrnet.git.

Keywords

Forward-looking sonar / Image denoising / Self-supervised learning / Wavelet transform

References

[1] Abdelhamed A, Lin S, Brown MS (2018) A high-quality denoising dataset for smartphone cameras. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Salt Lake City, pp 1692–1700

[2] Anwar S, Barnes N (2019) Real image denoising with feature attention. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, Seoul, pp 3155–3164

[3] Batson J, Royer L (2019) Noise2self: blind denoising by self-supervision. In: 36th International Conference on Machine Learning (ICML), Long Beach, pp 524–533

[4] Brummer B, De Vleeschouwer C (2019) Natural image noise dataset. In: 32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, pp 1712–1722

[5] Cho D, Bui TD. Multivariate statistical modeling for image denoising using wavelet transforms. Signal Proc-Image Commun, 2005, 20(1): 77-89,

[6] Dabov K, Foi A, Katkovnik V, Egiazarian K. Image denoising by sparse 3-D transform-domain collaborative filtering. IEEE Trans Image Proc, 2007, 16(8): 2080-2095,

[7] Guo S, Yan ZF, Zhang K, Zuo WM, Zhang L (2019) Toward convolutional blind denoising of real photographs. In: 32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, pp 1712–1722

[8] Guo TT, Mousavi HS, Vu TH, Monga V (2017) Deep wavelet prediction for image super-resolution. In: 30th IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, Honolulu, pp 104–113

[9] Gutmann M, Hyvärinen A (2010) Noise-contrastive estimation: a new estimation principle for unnormalized statistical models. In: 13th International Conference on Artificial Intelligence and Statistics (AISTATS), Sardinia, pp 297–304

[10] Hermans A, Beyer L, Leibe B (2017) In defense of the triplet loss for person reidentification. Preprint at arXiv:170307737. https://doi.org/10.48550/arXiv.1703.07737

[11] Huang T, Li SJ, Jia X, Lu HC, Liu JZ (2021) Neighbor2neighbor: self-supervised denoising from single noisy images. In: 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, pp, 14776–14785

[12] Jang G, Lee W, Son S, Lee K (2021) C2N: practical generative noise modeling for real-world denoising. In: 2021 IEEE/CVF International Conference on Computer Vision (ICCV), Montreal, pp 2350–2359

[13] Jensen JA, Svendsen NB. Calculation of pressure fields from arbitrarily shaped, apodized, and excited ultrasound transducers. IEEE Trans Ultraso Ferroelectr Freq Control, 1992, 39(2): 262-267,

[14] Krull A, Buchholz TO, Jug F (2019) Noise2void-learning denoising from single noisy images. In: 32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, pp 2124–2132

[15] Laine S, Karras T, Lehtinen J, Aila T, Wallach H, Larochelle H et al (2019) High-quality self-supervised deep image denoising. In: 33rd Conference on Neural Information Information Processing Systems (NeurIPS), Vancouver, pp 1–16

[16] Lee W, Son S, Lee KM (2022) AP-BSN: self-supervised denoising for real-world images via asymmetric PD and blind-spot network. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, New Orleans, pp 17704–17713

[17] Liu Y. Image denoising method based on threshold, wavelet transform and genetic algorithm. Int J Signal Proc Image Proc Pattern Recognit, 2015, 8(2): 29-40

[18] Mittal A, Moorthy AK, Bovik AC. No-reference image quality assessment in the spatial domain. IEEE Trans Image Proc, 2012, 21(12): 4695-4708,

[19] Neshatavar R, Yavartanoo M, Son S, Lee KM (2022) CVF-SID: cyclic multi-variate function for self-supervised image denoising by disentangling noise from image. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), New Orleans, pp 17562–17570

[20] Pang TY, Zheng H, Quan YH, Ji H (2021) Recorrupted-to-recorrupted: unsupervised deep learning for image denoising. In: IEEE/CVF Conference on Computer Vision and Pattern Recognition, Nashville, pp 2043–2052

[21] Park T, Efros AA, Zhang R, Zhu JY (2020) Contrastive learning for unpaired image-to-image translation. In: Computer Vision–ECCV 2020, Glasgow, pp 319–345

[22] Sheng YW, Xia ZG (1996) A comprehensive evaluation of filters for radar speckle suppression. In: IGARSS’96. 1996 International Geoscience and Remote Sensing Symposium, Lincoln, pp 1559–1561

[23] Simonyan K, Zisserman A (2014) Very deep convolutional networks for large-scale image recognition. Preprint at arXiv:14091556. https://doi.org/10.48550/arXiv.1409.1556

[24] Sohn K (2016) Improved deep metric learning with multi-class N-pair loss objective. In: 30th Conference on Neural Information Processing Systems (NIPS), Barcelona, pp 1857–1865

[25] Wang ZC, Fu Y, Liu J, Zhang YL (2023) LG-BPN: local and global blind-patch network for self-supervised real-world denoising. In: 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition, Vancouver, pp 18156–18165

[26] Wu XH, Liu M, Cao Y, Ren DW, Zuo WM, et al.. Vedaldi A, et al.. Unpaired learning of deep image denoising. Computer vision-ECCV 2020, 2020 Cham Springer 352-368,

[27] Xu B, Cui Y, Li ZH, Yang J. An iterative SAR image filtering method using nonlocal sparse model. IEEE Geosci Remote Sens Lett, 2015, 12(8): 1635-1639,

[28] Xu J, Huang Y, Cheng MM, Liu L, Zhu F, Xu Z, et al.. Noisy-as-Clean: learning self-supervised denoising from corrupted image. IEEE Trans Image Proc, 2020, 29: 9316-9329,

[29] Yang H, Wang Y. An effective and comprehensive image super resolution algorithm combined with a novel convolutional neural network and wavelet transform. IEEE Access, 2021, 9: 98790-98799,

[30] Yu S, Park B, Jeong J (2019) Deep iterative down-up CNN for image denoising. In: 32nd IEEE/CVF Conference on Computer Vision and Pattern Recognition, Long Beach, pp 2095–2013

[31] Zhang K, Li YW, Liang JY, Cao JZ, Zhang YL, Tang H, et al.. Practical blind image denoising via Swin-Conv-UNet and data synthesis. Mach Intell Res, 2023, 20(6): 822-836,

[32] Zhang K, Zuo WM, Chen YJ, Meng DY, Zhang L. Beyond a gaussian denoiser: residual learning of deep CNN for image denoising. IEEE Trans Image Proc, 2017, 26(7): 3142-3155,

[33] Zhang K, Zuo WM, Zhang L. FFDNet: toward a fast and flexible solution for CNN-based image denoising. IEEE Trans Image Proc, 2018, 27(9): 4608-4622,

[34] Zhou YQ, Jiao JB, Huang HB, Wang Y, Wang J, Shi HH et al (2020) When AWGN-based denoiser meets real noises. In: 34th AAAI Conference on Artificial Intelligence, New York, pp 13074–13081

Funding

Natural Science Foundation of Shandong Province(ZR2023MF012)

Sanya Yazhou Bay Science and Technology City(2021JJLH0061)

the Natural Science Foundation of China(41927805, 41906177)