基于信号匹配和最优分解层的小波去噪方法研究

发布时间：2018-05-01 23:55

本文选题：信号去噪 + 滤波器　；参考：《扬州大学》2014年硕士论文

【摘要】：在信号采集、传输和处理过程中,不可避免会含有噪声,直接影响后续处理的结果,因此如何对信号进行去噪处理,是目前信号处理领域中的国内外研究热点问题之一。目前关于信号去噪的方法很多,常用的去噪方法有基于傅立叶变换的信号去噪,由于傅立叶变换的时频单一性,信号的去噪效果较差。小波变换具有时域局部化特征、多分辨率特性、解相关特性和选基灵活性等特征,已广泛应用于雷达信号处理、语音识别、数据压缩、信号处理、模式识别、信号去噪等领域。针对不同的含噪信号,对七种常规小波基的去噪性能进行分析与实验比较,构造出基于结构化的9/7小波滤波器组和7/13小波滤波器组,并对其滤波器组进行性能分析比较。不同的小波基具有不同的时频特性,选择的小波基不同,对应的去噪的效果也不相同,所以在去噪过程中,小波基的选择,会直接影响小波去噪的效果,如何根据信号的特点选择最佳小波基,是目前国内外学者的研究的关键问题。针对现有小波基的不足,提出一种基于信号匹配的最优小波去噪方法。该方法根据信号在尺度空间的最大投影而构造的能量匹配准则,利用结构化小波滤波器组,结合遗传算法,构造出与信号能量一致的最优能量匹配小波。并根据波形匹配准则,结合结构化小波滤波器组,利用优化函数构造出与信号波形一致的最优波形匹配小波。实验确定结果表明,基于信号匹配的最优小波去噪方法的去噪效果优于其它小波。针对不同的信号去噪,除对不同小波基研究与最优小波基选取外,还要对基于小波分解层数的信号去噪进行了研究,发现小波分解层数与含噪信号的受污染程度存在一定关系。在实际的信号小波去噪过程中,不同的含噪信号,其小波分解层数不固定,且不同分解层数会对去噪效果产生很大的影响,针对这一问题,提出一种基于最优分解层去噪方法,该方法利用各个小波分解层的能量关系,即信噪比,结合优化算法来确定最优分解层进行去噪。实验确定结果表明,在它的最优分解层上的消噪效果达到了最佳。
[Abstract]:In the process of signal acquisition, transmission and processing, it is inevitable that there will be noise, which directly affects the results of subsequent processing. Therefore, how to Denoise the signal is one of the hot issues in the field of signal processing at home and abroad. At present, there are many methods of signal de-noising. The common methods of de-noising are based on Fourier transform. Because of the singularity of time-frequency of Fourier transform, the effect of signal de-noising is poor. Wavelet transform has been widely used in radar signal processing, speech recognition, data compression, signal processing, pattern recognition, signal denoising and so on. For different noisy signals, the performance of seven kinds of conventional wavelet bases is analyzed and compared by experiments. A structured 9 / 7 wavelet filter bank and a 7 / 13 wavelet filter bank are constructed, and the performance of the filter banks is analyzed and compared. Different wavelet bases have different time-frequency characteristics. Different wavelet bases have different corresponding denoising effects, so the selection of wavelet bases will directly affect the effect of wavelet denoising in the process of de-noising. How to select the best wavelet basis according to the characteristics of signal is the key problem of scholars at home and abroad. An optimal wavelet denoising method based on signal matching is proposed to overcome the shortcomings of existing wavelet bases. Based on the energy matching criterion constructed by the maximum projection of the signal in the scale space, the optimal energy matching wavelet which is consistent with the signal energy is constructed by using the structured wavelet filter bank and the genetic algorithm. According to the waveform matching criterion and the structured wavelet filter bank, the optimal waveform matching wavelet is constructed by using the optimization function. The experimental results show that the denoising effect of the optimal wavelet denoising method based on signal matching is better than that of other wavelets. For different signal denoising, besides the study of different wavelet bases and the selection of optimal wavelet bases, the signal denoising based on wavelet decomposition layers is also studied. It is found that the number of wavelet decomposition layers is related to the degree of contamination of noisy signals. In the actual signal wavelet denoising process, the wavelet decomposition layer number of different noisy signal is not fixed, and the different decomposition layer number will have a great influence on the de-noising effect. In view of this problem, a denoising method based on the optimal decomposition layer is proposed. In this method, the energy relationship of each wavelet decomposition layer, i.e. SNR, is used to determine the optimal decomposition layer for denoising. The experimental results show that the denoising effect on the optimal decomposition layer is the best.
【学位授予单位】：扬州大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN911.4

【参考文献】