|
|
|
| An Improved Multiple Threshold Wavelet Packet De-noising Algorithm and Its Application |
| WANG Hong-bin,WANG Shi-hao,JI Bing-shuo,ZHANG Hang-fei,QIAO Yong-jing,Xu Jian-tao |
| Institute of Electrical Engineering, Yanshan University, Qinhuangdao, Hebei 066004, China |
|
|
|
|
Abstract An improved multiple threshold wavelet packet de-noising algorithm is proposed, solving the problems of eliminating the noise incompletely and removing some useful signals without distinction. It was applied to the intelligent traffic image and got rid of the image signal de-noising incompletely and wrongly. First of all, the image was preprocessed using the decomposition reconstruction's algorithm of wavelet packet, and got more edge details. Then corresponding to different frequencies, threshold was set reasonable according to the characteristics of different energy adaptively, and different threshold was used to remove noise under different frequencies. Experiments showed that the method could remove the single noise effectively while preserving the image edges and details.
|
|
Received: 04 November 2014
Published: 27 January 2016
|
|
|
|
|
|
［1］王芳,季忠,彭承琳. 基于双树复小波变换的心电信号去噪研究［J］. 仪器仪表学报,2013,34(5):1160-1166.
［2］范秋凤,翟亚芳,周伟东.基于小波和小波包图像去噪的比较研究［J］.中国西部科技,2007,(15):19-21 .
［3］谢杰成,张大力,徐文立.小波图像去噪综述［J］.中国图像图形学报,2002,7(3):209-217.
［4］左平,王洋,申延成.一种基于全变差正则化与小波包变换的图像去噪算法［J］. 吉林大学学报（理学版）,2014,52(1):81-85.
［5］戴桂平,刘彬.基于小波去噪和 EMD 的信号瞬时参数提取［J］.计量学报,2007,28(2):158-162.
［6］Donoho D L. De-noising by Soft-thresholding［J］. IEEE Trans on Information Theory, 1995, 41(3): 613-627.
［7］孙健,王成华,杜庆波.基于小波包能量谱和NPE的模拟电路故障诊断［J］.仪器仪表学报,2013,34(9):2021-2027.
［8］孙勇,景博,覃征,等. 基于小波分析的信噪分离方法研究［J］.计量学报,2006,27(2):153-155.
［9］Stefano A,White P, Collis W.Training methods for image noise level estimation on wavelet components. ［J］. Eurasip Journal on Applied Signal Processing, 2004,16(6):2400-2407.
［10］陈世伟,李世平,管京周,等.图像融合技术在自动化测量中的应用［J］. 计量技术,2005,(11):12-14.
［11］张德丰.MATLAB 小波分析（2版）［M］.北京:机械工业出版社,2011.
［12］吴乐南. 数据压缩的原理与应用［ M］ . 北京 :电子工业出版社 ,1994. |
|
|
|
2016, Vol. 37 
