基于二维小波变换和独立分量分析的SAR图像去噪方法

卢晓光; 韩萍; 吴仁彪; 刘瑞华

doi:10.3724/SP.J.1146.2006.01668

基于二维小波变换和独立分量分析的SAR图像去噪方法

doi: 10.3724/SP.J.1146.2006.01668

基金项目:

国家杰出青年基金项目(60325102)，国家863计划(2006AA12Z313)，深圳大学ATR国防科技重点实验室开放课题(200603)和中国民航大学科研启动基金(Qd04q10)资助课题

计量
- 文章访问数: 3517
- HTML全文浏览量: 118
- PDF下载量: 1465
- 被引次数: 0
出版历程
- 收稿日期: 2006-10-30
- 修回日期: 2007-04-06
- 刊出日期: 2008-05-19

An Approach for SAR Image Despeckling Based on 2D-Wavelet Transform and ICA

摘要

摘要: 该文给出了一种基于离散小波变换和独立分量分析的SAR图像斑点噪声抑制方法。首先利用小波变换对图像进行分解,然后将分解出的各部分子图像分别进行独立分量分析,提取出相应的独立源,去除噪声分量,最后依次进行ICA重构和小波重构。该文还同时比较了采用不同小波基函数时斑点噪声的抑制效果,研究了它们对斑点抑制的影响。对MSTAR实测SAR图像的实验结果表明该方法能够有效地抑制图像中的斑点噪声,且在性能上优于ICA和Lee滤波方法。
- 合成孔径雷达; 斑点噪声抑制; 小波变换; 独立分量分析
Abstract: In this paper, a novel speckle suppression method is proposed, which is based on the joint of discrete wavelet transform and independent component analysis (ICA). This method employs ICA to process the wavelet coefficients of images, extracting the independent components which represent the uncontaminated images. Also this paper compares the suppressing performance of several standard wavelet bases，investigating their influences on suppressing the speckle. Experimental results based on SAR images of MSTAR dataset show that the proposed method suppresses speckle effectively, and outperforms ICA and Lee filter.

HTML全文

参考文献(1)

Lee J S. Speckle analysis and smoothing algorithm forsynthetic aperture radar images [J].Computer Graphic andImage Processing.1981, 17(1):24-32[2]Frost V S, Stiles J A, and Shanmugan K S, et al.. A model forradar images and its application to adaptive digital filteringof multiplicative noise [J].IEEE Trans. on Pattern Analysisand Machine Intelligence.1982, PAMI-4(2):157-166[3]Zhang J, Cheng X G, and Liu J. A speckle reductionalgorithm by soft-thresholding based on wavelet filters forSAR images [C]. Proc. Int. Conf. Signal Processing, Beijing,1998, 2: 1469-1472.[4]Xie H, Pierce L E, and Ulaby F T. SAR speckle reductionusing wavelet denoising and Markov random field modeling[J].IEEE Trans. on Geoscience and Remote Sensing.2002,40(10):2196-2212[5]Achim A, Tsakalides P, and Bezerianos A. SAR imagedenoising via Bayesian wavelet shrinkage based onheavy-tailed modeling [J].IEEE Trans. on Geoscience andRemote Sensing.2003, 41(8):1773-1784[6]Field D J. What is the goal of sensory coding [J]. NeuralComputation, 1994, 6(4): 559-601.[7]Hyv鋜inen A, Oja E, and Hoyer P, et al.. Image featureextraction by sparse coding and independent componentanalysis [C]. Proc. Int. Conf. on Pattern Recognition(ICPR'98), Brisbane, Australia, 1998, 12: 1268-1273.[8]Chen C H and Wang X J. A novel theory of SAR imagerestoration and enhancement with ICA [C]. IEEE Geoscienceand Remote Sensing Symposium, Anchorage, America, 2004,6: 3911-3914.[9]Hyv鋜inen A. Fast and robust fixed-point algorithms forindependent component analysis [J]. IEEE Trans. on NeuralNetworks, 1999, 10(3): 626-634.[10]Bell A J and Sejnowski T J. An Information-Maximizationapproach to blind separation and blind deconvolution [J].Neural Computation.1995, 7(6):1129-1159[11]Mallat S G. A theory for multiresolution signaldecomposition: The wavelet representation [J].IEEE Trans.on Pattern Analysis and Machine Intelligence.1989, 11(7):674-693[12]Daubechies I. Ten Lectures on Wavelets. CBMS-NSFRegional Conf. Series in Applied Mathematics, PhiladelphiaPennsylvania, March 1992, 61: 67-72.[13]Azzerboni B, Finocchio G, and Ipsale M, et al.. A newapproach to detection of muscle activation by independentcomponent analysis and wavelet transform. Springer-Verlag,Lecture Notes in Computer Science, 2002, 2486: 109-116.[14]Azzerboni B, Carpentieri M, and La Foresta F, et al..Neural-ICA and wavelet transform for artifacts removal insurface EMG [J]. Proceedings of 2004 International JointConf. on Neural Networks, Budapest, Hungary, 2004, 4:3223-3228.[15]Oliver C and Quegan S. Understanding Synthetic ApertureRadar Images [M]. Boston: Artech House, 1998, Part 4.4.

施引文献

资源附件(0)

访问统计