政大機構典藏-National Chengchi University Institutional Repository(NCCUR):Item 140.119/157813

English | 正體中文 | 简体中文 | Post-Print筆數 : 27 | Items with full text/Total items : 116849/147881 (79%)
Visitors : 64181992 Online Users : 658

RC Version 6.0 © Powered By DSPACE, MIT. Enhanced by NTU Library IR team.

Scope

please add "double quotation mark" for query phrases to get precise results

please goto advance search for comprehansive author search

Adv. Search

Home ‧ Login ‧ Upload ‧ Help ‧ About ‧ Administer

Goto mobile version

政大機構典藏 > 資訊學院 > 資訊科學系 > 學位論文 > Item 140.119/157813

Please use this identifier to cite or link to this item: https://nccur.lib.nccu.edu.tw/handle/140.119/157813

Title:	透過最小-最大均值池化濾波器與多頭自注意力卷積神經網路之影像雜訊移除研究 Image Denoising Using Min-Max Mean Pooling Filters and Multi-Head Self-Attention Convolutional Neural Networks
Authors:	陸妍諭 Lu, Yen-Yu
Contributors:	張宏慶 Jang, Hung-Chin 陸妍諭 Lu, Yen-Yu
Keywords:	多頭自注意力神經網路卷積神經網路深度學習均值濾波器椒鹽雜訊池化濾波 multi-head self-attention neural network convolutional neural network deep learning mean filter salt-and-pepper noise pooling-based filtering
Date:	2025
Issue Date:	2025-07-01 15:06:44 (UTC+8)
Abstract:	數位影像在傳輸的過程中，可能會受到電磁干擾和攝影元件受損，導致影像受到脈衝雜訊干擾而損壞，如何有效修復遭受極值脈衝雜訊(椒鹽雜訊)干擾的影像，對於提升數位影像品質極為重要。本文提出一種使用自適應分析視窗最小-最大均值池化的多頭自注意力卷積神經網路，去除在傳輸過程中產生的椒鹽雜訊；首先，估測影像的雜訊密度，若為輕度雜訊干擾之影像，乾淨像素足夠，使用多頭自注意力卷積神經網路，計算輸入序列中不同位置像素的權重，並擷取長距離像素間的相關性，預測適合重建的臨域像素；相對的，若是中、高雜訊密度干擾時，受損像素較多，則透過多層的最小值和最大值均值池化濾波器，分別計算最大池化後的影像和最小池化後的兩類影像；最後將處理後的影像重新組合，並進行均值濾波處理；在高雜訊密度的環境中，若分析視窗沒有乾淨像素可以參考，則擴大分析視窗的尺寸，增加鄰域未受干擾像素引入的機率，修復影像中的受損像素。經由實驗結果證明：本文所提出的濾波器可以在各種雜訊密度有效重建受到雜訊干擾的影像，重建效果也優於許多極為先進(state-of-the-art)的演算法。 During the transmission of digital images, electromagnetic interference and sensor defects can result in impulse noise corruption, leading to severe image degradation. Effectively restoring images contaminated by extreme impulse noise, such as salt-and-pepper noise, is crucial for improving digital image quality. This paper proposes a multi-head self-attention neural network enhanced by adaptive min-max mean pooling to remove salt-and-pepper noise introduced during transmission. First, the noise density of the corrupted image is estimated. Suppose the image is subjected to low-level noise and contains a sufficient number of clean pixels; a multi-head self-attention mechanism is employed to compute the attention weights of pixels at different positions in the input sequence. This mechanism captures long-range dependencies to predict suitable neighborhood pixels for reconstruction. In contrast, under moderate to high noise density, where a larger portion of pixels are corrupted, the method utilizes multiple layers of min and max mean pooling filters to compute two feature maps separately based on maximum and minimum pooling operations. The processed images are then fused and further refined using mean filtering. In high-noise scenarios where no clean pixels are available within a small analysis window, the window size is adaptively enlarged to increase the likelihood of including undisturbed neighboring pixels, thereby improving the reconstruction of damaged regions. Experimental results demonstrate that the proposed filtering method can effectively restore noise-corrupted images across various noise levels, outperforming several state-of-the-art algorithms regarding reconstruction quality.
Reference:	[1] R. C. Gonzalez, R. E. Woods, and S. L. Eddins, Digital Image Processing, 2nd ed. New York, NY, USA: Prentice-Hall, 2002. [2] J. D. Gibson and A. Bovik, Handbook of Image and Video Processing, 1st ed. San Diego, CA, USA: Academic Press, 2000. [3] H. Hwang and R. A. Haddad, "Adaptive median filters: new algorithms and results," IEEE Trans. Image Process., vol. 4, no. 4, pp. 499–502, Apr. 1995. [4] S.-J. Ko and Y. H. Lee, "Center weighted median filters and their applications to image enhancement," IEEE Trans. Circuits Syst., vol. 38, no. 9, pp. 984–993, Sep. 1991. [5] Y. Dong and S. Xu, "A new directional weighted median filter for removal of random-valued impulse noise," IEEE Signal Process. Lett., vol. 14, no. 3, pp. 193–196, Mar. 2007. [6] T. Chen and H. R. Wu, "Adaptive impulse detection using center-weighted median filters," IEEE Signal Process. Lett., vol. 8, no. 1, pp. 1–3, Jan. 2001. [7] C.-T. Lu and T.-C. Chou, "Denoising of salt-and-pepper noise corrupted image using modified directional-weighted-median filter," Pattern Recognit. Lett., vol. 33, no. 10, pp. 1287–1295, Jul. 2012. [8] S. Esakkirajan, T. Veerakumar, A. N. Subramanyam, and C. H. PremChand, "Removal of high density salt and pepper noise through modified decision based unsymmetric trimmed median filter," IEEE Signal Process. Lett., vol. 18, no. 5, pp. 287–290, May 2011. [9] T. Chen, K.-K. Ma, and L.-H. Chen, "Tri-state median filter for image denoising," IEEE Trans. Image Process., vol. 8, no. 12, pp. 1834–1838, Dec. 1999. [10] P. Satti, N. Sharma, and B. Garg, "Min-max average pooling based filter for impulse noise removal," IEEE Signal Process. Lett., vol. 27, pp. 1475–1479, 2020. [11] Y.-Y. Lu and H.-C. Jang, "Removal of salt-and-pepper impulse noise using adaptive analysis windows with min-max mean pooling," in Proc. 36th IPPR Conf. Comput. Vis. Graph. Image Process., 2023. [12] L. Zhai, L. Dong, Y. Liu, S. Fu, F. Wang, and Y. Li, "Adaptive hybrid threshold shrinkage using singular value decomposition," J. Electron. Imaging, vol. 28, no. 6, pp. 063002, Nov. 2019. [13] S. H. Chan, T. Zickler, and Y. M. Lu, "Monte Carlo non-local means: Random sampling for large-scale image filtering," IEEE Trans. Image Process., vol. 23, no. 8, pp. 3711–3725, Aug. 2014. [14] E. Luo, S. H. Chan, and T. Q. Nguyen, "Adaptive image denoising by mixture adaptation," IEEE Trans. Image Process., vol. 25, no. 10, pp. 4489–4503, Oct. 2016. [15] S. Anwar, F. Porikli, and C. P. Huynh, "Category-specific object image denoising," IEEE Trans. Image Process., vol. 26, no. 11, pp. 5506–5518, Nov. 2017. [16] J. Xu, L. Zhang, and D. Zhang, "External prior guided internal prior learning for real-world noisy image denoising," IEEE Trans. Image Process., vol. 27, no. 6, pp. 2996–3010, Jun. 2018. [17] M. S. M. Zain, A. K. Junoh, and A. Abdurrazzaq, "Hybrid singular value decomposition based alpha trimmed mean-median filter in eliminating high density salt and pepper noise from grayscale image," Multimed. Tools Appl., vol. 83, pp. 62895–62913, 2024. [18] J. Gao, L. Li, X. Ren, Q. Chen, and Y. M. Abdul-Abbass, "An effective method for salt and pepper noise removal based on algebra and fuzzy logic function," Multimed. Tools Appl., vol. 83, pp. 9547–9576, 2024. [19] Q. Liu, X. Li, and J. Yang, "Optimum codesign for image denoising between type-2 fuzzy identifier and matrix completion denoiser," IEEE Trans. Fuzzy Syst., vol. 30, no. 1, pp. 287–292, Jan. 2022. [20] X. Zhang, X. Li, Z. Tang, S. Zhang, and S. Xie, "Noise removal in embedded image with bit approximation," IEEE Trans. Knowl. Data Eng., vol. 34, no. 3, pp. 1359–1369, Mar. 2022. [21] N. N. Hien, D. N. H. Thanh, U. Erkan, and J. M. R. S. Tavares, "Image noise removal method based on thresholding and regularization techniques," IEEE Access, vol. 10, pp. 71584–71597, 2022. [22] N. Aslam, M. K. Ehsan, Z. U. Rehman, M. Hanif, and G. Mustafa, "A modified form of different applied median filter for removal of salt & pepper noise," Multimed. Tools Appl., vol. 82, pp. 7479–7490, 2023. [23] S. Zhang and M. A. Karim, "A new impulse detector for switching median filters," IEEE Signal Process. Lett., vol. 9, no. 11, pp. 360–363, Nov. 2002. [24] C.-T. Lu, Y.-Y. Chen, L.-L. Wang, and C.-F. Chang, "Removal of salt-and-pepper noise in corrupted image using three-values-weighted approach with variable-size window," Pattern Recognit. Lett., vol. 80, pp. 188–199, Sep. 2016. [25] D. Liu, B. Wen, J. Jiao, X. Liu, Z. Wang, and T. S. Huang, "Connecting image denoising and high-level vision tasks via deep learning," IEEE Trans. Image Process., vol. 29, pp. 3695–3706, 2020. [26] Z. Liu, W. Q. Ya, and M. L. Yang, "Image denoising based on a CNN model," in Proc. Int. Conf. Control, Autom. Robot., 2018, pp. 389–393. [27] A. V. Miclea, R. Terebes, I. Ilea, and M. Borda, "Hyperspectral image classification using combined spectral-spatial denoising and deep learning techniques," in Proc. IEEE Int. Conf. Autom., Qual. Testing, Robot. (AQTR), 2018. [28] K. Zhang, W. Zuo, Y. Chen, D. Meng, and L. Zhang, "Beyond a Gaussian denoiser: Residual learning of deep CNN for image denoising," IEEE Trans. Image Process., vol. 26, no. 7, pp. 3142–3155, Jul. 2017. [29] X. Li, Q. Gao, Y. Lu, and D. Sun, "Convolutional neural network denoising method based on multisize features," in Proc. IEEE Int. Conf. Intell. Human-Machine Syst. Cybern., 2018, pp. 146–149. [30] K. Zhang, W. Zuo, and L. Zhang, "FFDNet: Toward a fast and flexible solution for CNN-based image denoising," IEEE Trans. Image Process., vol. 27, no. 9, pp. 4608–4622, Sep. 2018. [31] T. Wang, M. Sun, and K. Hu, "Dilated deep residual network for image denoising," in Proc. IEEE Int. Conf. Tools Artif. Intell., 2017, pp. 1272–1279. [32] K. Isogawa, T. Shiodera, and T. Takeguchi, "Deep shrinkage convolutional neural network for adaptive noise reduction," IEEE Signal Process. Lett., vol. 25, no. 2, pp. 224–228, Feb. 2018. [33] E. Kandić, A. Akagic, and M. Bohlouli, "Exploring convolutional autoencoder efficacy in noise removal for image processing and computer vision: A study using the MNIST dataset," in Proc. 10th Int. Conf. Control, Decision Inf. Technol. (CoDIT), Vallette, Malta, 2024, pp. 1275–1280. [34] L. Deng, "The MNIST database of handwritten digit images for machine learning research," IEEE Signal Process. Mag., vol. 29, no. 6, pp. 141–142, Nov. 2012. [35] C.-T. Lu, R.-H. Chen, L.-L. Wang, and J.-A. Lin, "Image enhancement using convolutional neural network to identify similar patterns," IET Image Process., vol. 14, no. 17, pp. 3880–3889, 2020. [36] C.-T. Lu, H.-J. Hsu, and L.-L. Wang, "Image denoising using DLNN to recognize the direction of pixel variation," Signal Image Video Process., vol. 15, pp. 1247–1256, 2021. [37] A. A. Rafiee and M. Farhang, "A deep convolutional neural network for salt-and-pepper noise removal using selective convolutional blocks," Appl. Soft Comput., vol. 145, 2023. [38] C.-T. Lu, L.-L. Wang, J.-H. Shen, and J. A. Lin, "Image enhancement using deep-learning fully connected neural network mean filter," J. Supercomput., vol. 77, pp. 3144–3164, 2021. [39] Q. Yang, P. Yan, M. K. Kalra, and G. Wang, "CT image denoising with perceptive deep neural networks," arXiv preprint, arXiv:1702.07019, 2017. [40] K. Simonyan and A. Zisserman, "Very deep convolutional networks for large-scale image recognition," arXiv preprint arXiv:1409.1556, 2014. [41] E. Kang and J. C. Ye, "Framelet denoising for low-dose CT using deep learning," in Proc. IEEE Int. Symp. Biomed. Imaging (ISBI), 2018, pp. 311–314. [42] H. Yuan, J. Jia, and Z. Zhu, "SIPID: A deep learning framework for sinogram interpolation and image denoising in low-dose CT reconstruction," in Proc. IEEE Int. Symp. Biomed. Imaging (ISBI), 2018, pp. 1521–1524. [43] Q. Yang, P. Yan, Y. Zhang, H. Yu, Y. Shi, X. Mou, M. K. Kalra, Y. Zhang, L. Sun, and G. Wang, "Low-dose CT image denoising using a generative adversarial network with Wasserstein distance and perceptual loss," IEEE Trans. Med. Imaging, vol. 37, no. 6, pp. 1348–1357, Jun. 2018. [44] M. A. Marnissi, "Revolutionizing thermal imaging: GAN-based vision transformers for image enhancement," in Proc. IEEE Int. Conf. Image Process. (ICIP), Kuala Lumpur, Malaysia, 2023, pp. 2735–2739. [45] S. I. Jang, T. Pan, Y. Li, P. Heidari, and J. Chen, "Spach Transformer: Spatial and channel-wise transformer based on local and global self-attentions for PET image denoising," IEEE Trans. Med. Imaging, vol. 43, no. 6, pp. 2036–2049, Jun. 2024. [46] Z. Li, J. Zhang, S. Wei, Y. Gao, C. Cao, and Z. Wu, "TPAFNet: Transformer-driven pyramid attention fusion network for 3D medical image segmentation," IEEE J. Biomed. Health Inform., vol. 28, no. 11, pp. 6803–6814, Nov. 2024. [47] M. A. Marnissi and A. Fathallah, "GAN-based vision transformer for high-quality thermal image enhancement," in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. Workshops (CVPRW), Vancouver, BC, Canada, 2023, pp. 817–825. [48] Z. Gao, Z. Tong, K. Q. Lin, J. Chen, and M. Z. Shou, "Bootstrapping sparseformers from vision foundation models," in Proc. IEEE/CVF Conf. Comput. Vis. Pattern Recognit. (CVPR), Seattle, WA, USA, 2024, pp. 17710–17721. [49] S. Lu, W. Zhang, H. Zhao, H. Liu, N. Wang, and H. Li, "Anomaly detection for medical images using heterogeneous auto-encoder," IEEE Trans. Image Process., vol. 33, pp. 2770–2782, 2024. [50] M. S. I. Sajol and A. S. M. J. Hasan, "Benchmarking CNN and cutting-edge transformer models for brain tumor classification through transfer learning," in Proc. IEEE 12th Int. Conf. Intell. Syst. (IS), Varna, Bulgaria, 2024, pp. 1–6. [51] M. Songade, S. Arya, and S. M. Moorthi, "Single band NIR-to-RGB image colorization using attention guided conditional GAN," in Proc. IEEE Space, Aerosp. Defence Conf. (SPACE), Bangalore, India, 2024, pp. 407–410. [52] M. H. Kolekar, S. Bose, and A. Pai, "SARain-GAN: Spatial attention residual UNet based conditional generative adversarial network for rain streak removal," IEEE Access, vol. 12, pp. 43874–43888, 2024. [53] D. Zhang, N. Tang, and Y. Qu, "Joint motion deblurring and super-resolution for single image using diffusion model and GAN," IEEE Signal Process. Lett., vol. 31, pp. 736–740, 2024. [54] D. Martin, C. Fowlkes, D. Tal, and J. Malik, "A database of human segmented natural images and its application to evaluating segmentation algorithms and measuring ecological statistics," in Proc. 8th Int. Conf. Comput. Vis., vol. 2, Jul. 2001, pp. 416–423. [55] A. Paszke, S. Gross, F. Massa, A. Lerer, J. Bradbury, G. Chanan, T. Killeen, Z. Lin, N. Gimelshein, L. Antiga, et al., "PyTorch: An imperative style, high-performance deep learning library," in Advances in Neural Information Processing Systems, vol. 32, 2019, pp. 8026–8037.
Description:	碩士國立政治大學資訊科學系 112753205
Source URI:	http://thesis.lib.nccu.edu.tw/record/#G0112753205
Data Type:	thesis
Appears in Collections:	[資訊科學系] 學位論文

Files in This Item:

File	Size	Format
320501.pdf	12422Kb	Adobe PDF	0	View/Open

All items in 政大典藏 are protected by copyright, with all rights reserved.

社群 sharing

著作權政策宣告 Copyright Announcement

1.本網站之數位內容為國立政治大學所收錄之機構典藏，無償提供學術研究與公眾教育等公益性使用，惟仍請適度，合理使用本網站之內容，以尊重著作權人之權益。商業上之利用，則請先取得著作權人之授權。
The digital content of this website is part of National Chengchi University Institutional Repository. It provides free access to academic research and public education for non-commercial use. Please utilize it in a proper and reasonable manner and respect the rights of copyright owners. For commercial use, please obtain authorization from the copyright owner in advance.

2.本網站之製作，已盡力防止侵害著作權人之權益，如仍發現本網站之數位內容有侵害著作權人權益情事者，請權利人通知本網站維護人員(nccur@nccu.edu.tw)，維護人員將立即採取移除該數位著作等補救措施。
NCCU Institutional Repository is made to protect the interests of copyright owners. If you believe that any material on the website infringes copyright, please contact our staff(nccur@nccu.edu.tw). We will remove the work from the repository and investigate your claim.

DSpace Software Copyright © 2002-2004 MIT & Hewlett-Packard / Enhanced by NTU Library IR team Copyright © - Feedback