基于K-SVD的低信噪比WMSN视频图像稀疏去噪

J4 ›› 2014, Vol. 36 ›› Issue (03): 497-501.

基于K-SVD的低信噪比WMSN视频图像稀疏去噪

罗晖，褚红亮，王世昌

(华东交通大学信息工程学院,江西南昌 330013)

收稿日期:2012-10-15 修回日期:2012-12-22 出版日期:2014-03-25 发布日期:2014-03-25
基金资助:
国家自然科学基金资助项目（61261040）

K-SVD based sparse denoising for
WMSN video image with low SNR

LUO Hui,CHU Hongliang,WANG Shichang

(School of Information Engineering,East China Jiaotong University,Nanchang 330013,China)

Received:2012-10-15 Revised:2012-12-22 Online:2014-03-25 Published:2014-03-25

摘要/Abstract

摘要：

无线多媒体传感器网络WMSN因感知视频等信息的优势而被广泛应用，但受天气、光照等外因干扰，所采集视频图像常含有较为严重的噪声。因此，在低信噪比条件下进行视频图像去噪是保证WMSN视频监测有效性和可靠性的关键。在分析WMSN视频图像特征的基础上，首先对其进行周期性采集、分帧及帧差等预处理；然后对关键帧运用KSVD训练DCT冗余字典以充分稀疏表示图像特征，并采用基于残差比的改进型BatchOMP实现关键帧去噪及重构，而对残差帧则基于DCT冗余字典进行稀疏去噪处理；最后，叠加去噪后的关键帧和残差帧，从而整体上实现低信噪比WMSN视频图像的去噪及重构。实验表明，本算法能更加有效地、较为快速地滤除视频图像噪声，适用于低信噪比WMSN视频图像去噪。

关键词: 稀疏去噪, K奇异值分解, 残差比, 低信噪比, 无线多媒体传感器网络

Abstract:

As a highly effective method of perceiving multimedia information, Wireless Multimedia Sensor Networks (WMSNs) has shown its potential in many areas. However, the outside interference in the monitoring environment brings severe noise to video images. Obviously, video image denoising becomes the key to ensuring the validity and reliability of WMSN video monitoring. To denoise WMSN video image, firstly, its features are analyzed and some pretreatment are done. Secondly, the KSVD algorithm is employed to adaptively train DCT dictionary for reflecting the image characteristics and reconstruct the key frame through improved BatchOMP algorithm with residual ratio as the iteration termination, while DCT dictionary is adopted to sparsely denoise the residual frames. Finally, the video image is reconstructed under the situation of low SNR. Experimental results show that, compared with its counterparts, the superiorities of the algorithm can be observed in both visual and some numerical guidelines, showing the suitability for the WMSN video image denoising in low SNR.

Key words: sparse denoising;K-SVD;residual ratio;low SNR;wireless multimedia sensor network

罗晖，褚红亮，王世昌. 基于K-SVD的低信噪比WMSN视频图像稀疏去噪[J]. J4, 2014, 36(03): 497-501.

LUO Hui,CHU Hongliang,WANG Shichang. K-SVD based sparse denoising for
WMSN video image with low SNR [J]. J4, 2014, 36(03): 497-501.

参考文献

［1］Lu Qin. Research on multinode cooperative image acquisition and compression technology in wireless multimedia sensor networks ［D］. Changsha:National University of Defense Technology, 2009. (in Chinese)
［2］Pinar S, Kerem I, Sebnem B, et al. Image quality estimation in wireless multimedia sensor networks:An experimental study［J］. BROADNETS, LNICST, 2012, 66(2):226241.
［3］Gribonval R, Nielsen M. Sparse approximations in signal and image processing ［J］. Signal Processing,2006, 86(3):415416.
［4］Yaghoobi M, Blumensath T, Davies M E. Dictionary learning for sparse approximations with the majorization method［J］. IEEE Transactions on Signal Processing, 2009, 57(6):21782191.
［5］Candes E, Donoho D. New tight frames of curvelets and optimal representations of objects with C2 singularities［J］. Common Pure and Applied Math, 2004, 57(2):219266.
［6］Aharon M, Elad M, Bruckstrein A. The KSVD:An algorithm for designing of overcomplete dictionaries for sparse representation ［J］. IEEE Transactions on Signal Processing, 2006, 54 (11):43114322.
［7］Aharon M, Elad M. Sparse and redundant modeling of image content using an imagesignaturedictionary ［J］. SIAM Journal on Imaging Sciences, 2008, 1(3):228247.
［8］Rubinstein R, Zibulevsky M, Elad M. Efficient implementation of the KSVD algorithm using batch orthogonal matching pursuit［R］. CS200808.Haifa, Israel:Computer Science Department, TechnionIsrael Institute of Technology, 2008.
［9］Garg K, Nayar S K. Vision and rain［J］. International Journal of Computer Vision, 2007, 75(1):327.
［10］Liang Wei, Que Peiwen, Chen Liang, et al. Residual ratio iteration termination condition for MP method［J］. Journal of Shanghai Jiaotong University, 2010, 44(2):171175. (in Chinese)
［11］Cunha L A, Zhou J P, Do M N. The nonsubsampled contourlet transform:Theory, design, and applications ［J］. IEEE Transactions on Image Processing, 2006, 15(10):30893101.
附中文参考文献：
［1］鲁琴. 无线多媒体传感器网络多节点协同图像采集与压缩技术研究［D］.长沙:国防科学技术大学, 2009.
［10］梁巍, 厥沛文, 陈亮, 等. 基于残差比阈值的迭代终止条件匹配追踪稀疏分解方法［J］.

[1]	王涛, 全海燕. 低信噪比下联合训练生成对抗网络的语音分离[J]. 计算机工程与科学, 2021, 43(06): 1088-1094.
[2]	陈显毅1,2，金志刚1，苏毅珊1. 无线多媒体传感器网络能量均衡的多跳图像传输机制[J]. 计算机工程与科学, 2017, 39(11): 2028-2036.
[3]	罗晖，刘洁丽，祁美丽. 基于稀疏表示的WMSN红外和可见光图像融合[J]. J4, 2014, 36(05): 957-962.
[4]	朱琳，张可，董海兰. 基于虚拟网格的干扰避免多路径路由协议[J]. J4, 2013, 35(1): 67-71.