Filter Bank Based Nonlocal Means for Denoising Magnetic Resonance Images

doi:10.1007/s12204-014-1476-8

Abstract

Abstract: Image denoising is a classical problem in image processing. Its essential goal is to preserve the image features and to reduce noise effectively. The nonlocal means (NL-means) filter is a successful approach proposed in recent years due to its patch similarity comparison. However, the accuracy of similarities in this algorithm degrades when it suffers from heavy noise. In this paper, we introduce feature similarities based on a multichannel filter into NL-means filter. The multi-bank based feature vectors of each pixel in the image are computed by convolving from various orientations and scales to Leung-Malik set (edge, bar and spot filters), and then the similarities based on this information are computed instead of pixel intensity. Experiments are carried out with Rician noise. The results demonstrate the superior performance of the proposed method. The wavelet-based method and traditional NL-means in term of both mean square error (MSE) and perceptual quality are compared with the proposed method, and structural similarity (SSIM) and quality index based on local variance (QILV) are given.

Key words: Leung-Malik (LM) filter| nonlocal means| similarity| image denoising| Rician noise

摘要： Image denoising is a classical problem in image processing. Its essential goal is to preserve the image features and to reduce noise effectively. The nonlocal means (NL-means) filter is a successful approach proposed in recent years due to its patch similarity comparison. However, the accuracy of similarities in this algorithm degrades when it suffers from heavy noise. In this paper, we introduce feature similarities based on a multichannel filter into NL-means filter. The multi-bank based feature vectors of each pixel in the image are computed by convolving from various orientations and scales to Leung-Malik set (edge, bar and spot filters), and then the similarities based on this information are computed instead of pixel intensity. Experiments are carried out with Rician noise. The results demonstrate the superior performance of the proposed method. The wavelet-based method and traditional NL-means in term of both mean square error (MSE) and perceptual quality are compared with the proposed method, and structural similarity (SSIM) and quality index based on local variance (QILV) are given.

关键词: Leung-Malik (LM) filter| nonlocal means| similarity| image denoising| Rician noise

CLC Number:

TN 911.73

GUO Tian-li (郭甜莉), LIU Qie-gen (刘且根), LUO Jian-hua* (骆建华). Filter Bank Based Nonlocal Means for Denoising Magnetic Resonance Images[J]. Journal of shanghai Jiaotong University (Science), 2014, 19(1): 72-78.

References

[1] Lysaker M, Lundervold A, Taj X C. Noise removal using fourth-order partial differential equations with applications to medical magnetic resonance images in space and time [J]. IEEE Transaction on Image Process, 2003, 12(12): 1579-1590.
[2] Gerig G, Kubler O, Kikinis R, et al. Nonlinear anisotropic filtering of MRI data [J]. IEEE Transactions on Medical Imaging, 1992, 11(2): 221-232.
[3] Perona P, Malik J. Scale space and edge detection using anisotropic diffusion [J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1990,12(7): 629-639.
[4] Buades A, Coll B, Morel J M. A review of image denoising algorithms, with a new one [J]. SIAM Journal on Multiscale Modeling and Simulation, 2005,4(2): 490-530.
[5] Efros A A, Leung T K. Texture synthesis by non parametric sampling [C]//IEEE International Conference on Computer Vision. Corfu, Greece: IEEE, 1999:1033-1038.
[6] Mahmoudi M, Sapiro G. Fast image and video denoising via nonlocal means of similar neighborhoods[J]. IEEE Transactions on Signal Process Letters, 2005,12: 839-842.
[7] Kervrann C, Boulanger J. Optimal spatial adaptation for patch based image denoising [J]. IEEE Transaction on Image Process, 2006, 15(10): 2866-2878.
[8] Coup′e P, Yger P, Prima S, et al. An optimized blockwise nonlocal means denoising filter for 3-D magnetic resonance images [J]. IEEE Transaction on Medical Imaging, 2008, 27(4): 425-441.
[9] Manj′on J V, Caballero-Caballero J, Lull J J,et al. MRI denoising using non-local means [J]. Medical Image Analysis, 2008, 12: 514-523.
[10] Wong A, Mishra A K. Quasi-Monte Carlo estimation approach to denoising MRI data based on regional statistics [J]. IEEE Transactions on Biomedical Engineering,2011, 58(4): 1076-1083.
[11] He L L, Greenshields I R. A nonlocal maximum likelihood estimation method for Rician noise reduction in MR images [J]. IEEE Transactions on Medical Imaging, 2009, 28(2): 165-172.
[12] Leung T, Malik J. Representing and recognizing the visual appearance of materials using three-dimensional textons [J]. International Journal of Computer Vision,2001, 43(1): 29-44.
[13] Collins D L, Zijdenbos A P, Kollokian V, et al.Evans A. Design and construction of a realistic digital brain phantom [J]. IEEE Transaction on Medical Imaging, 1998, 17(3): 463-468.
[14] Wang Z, Bovik A C, Sheikh H R, et al. Image quality assessment: From error visibility to structural similarity[J]. IEEE Transaction on Image Process, 2004,13(4): 600-612.
[15] Aja-Fern′andez S, Est′epar R S J, Alberola-L′opez C, et al. Image quality assessment based on local variance [C]//Proceedings of the 28th IEEE EMBS Annual International Conference. New York: IEEE,2006: 4815-4818.