给实变换中无级变阶或变宽镶边频响附加噪声化相谱
实变换,指Matlab的小波工具箱代表的离散小波变换,其中被处理信号和滤波器时域序列都取实数值。变换局限于实线性空间时,其滤波器组必须受到更多的约束。本短文给出,为Matlab所需的滤波器序列附加噪声化相谱的方面例子,希望更好地破除,基于无穷连续时间信号的变换思想的束缚。不管是否做实变换,都最好把低通滤波器的直流增益,标准化为根号2。
以分数阶幅频响应、无级变宽余弦镶边幅频响应为例,因为它们是新颖罕见的(《维基百科》中最新版“Wavelet”部分摘录于本文后),也超越了使用FIR滤波器的Matlab和Wavelab旧制。但是,这里的用法,也适用于多孔化等处理过的滤波器。试验程序mlt_DWT_FPf.m,如图片1.所示。
其中,处理分数阶频响的函数FPfFracorder,和处理无级变宽余弦镶边频响的函数FPfCosrim,分别对应于《用小波包变换测验双重噪声化相谱的多孔无级变阶频响》(2015-11-05)中的PfFracorder,和《用小波包变换检验无级变宽余弦镶边频响》(2015-08-14)中的PfCosrim。
它们的函数名,用大写字母F开头,表示与对应无F时函数的差别:两个输出,为周期化滤波器的频谱序列,未经过逆傅立叶变换,这可减少某些应用中反复的FFT和IFFT;第一个输出,对应低通滤波器,其直流增益已是标准的;第二个,对应高通滤波器,由直接在频域实现CQF处理而得,不再是空的。
把上述两个频域输出,分别与具有合适对称性随机相位和单位复模的序列的对应元素相乘,再由IFFT变换到时域。然后,仅取用第一个序列的实部,作为Matlab小波工具箱处理正交FIR滤波器组的函数orthfilt的输入,可得4个输出,即是,Matlab的离散小波正变换函数wavedec、逆变换函数waverec所用的“4个FIR滤波器”。
第二个序列(不需real取实部,不参与小波变换),与orthfilt输出中那个重建高通滤波器序列,可以相加或相减。在和序列、差序列的两个范数值当中,取较小者,记为误差e0,用于评估不同计算方式间的兼容性。这里分加、减两种情况,因为工具箱做常规FIR的qmf处理(如图片2.所示)时,默认的高通滤波器的符号,可能与离散傅立叶变换和CQF(参见名著“十讲”5.6节)默认时间零点规则不一致。
随机序列x,经wavedec分解和waverec重建后得y。把它们的差序列的范数,除以x的范数后,再加上前述兼容性误差e0,记入误差输出矩阵Er1。Er1中的列号,对应x的长度的指标;行号,对应滤波器编号。编号1至40,用分数阶频响,其阶参数在区间内随机取值;编号41至180,用无级变宽余弦镶边频响,其多项式阶数为1至8,其平顶宽度参数在 内随机取值。
试验程序mlt_DWT_FPf(mf0,dl0,ns0)的3个输入参数,依次为,测试信号的可取长度值的数目(10至512)、最小分解深度、随机数发生器种子。测试的信号长度为,指标jj乘以2的dl0次方。程序结束时,在图形窗中,画出纪录的4个滤波器序列(两类频响的分解低通、重建高通的例子)及其幅谱。
图片1.中右上角,命令窗里显示例子,Er1=mlt_DWT_FPf(256,1,88),证实了高精度正交变换。测试涵盖了2至512的所有偶数长度的信号,最大分解深度,已达9。
另外,人们可能习惯于,小波工具箱的降噪函数wden和常规离散小波变换,做无冗余的处理。如果,在试过居士的Tpwp时,在其合成函数的输入中,未强行添加不能归于同一组基的数据点,那么,也遇不到,程序因为未预定冗余合成规则而报告错误后终止运行的情形。冗余合成(《冗余小波包合成以及降噪试验和能量守恒定标》,2015-12-27),常被忽略。实际上,用wden、wavedec做L级分解的全过程,已涵盖了小波包树上的L+1组基,不过只留用了最后的一组基的数据。
将A经过一级分解后,得低通输出B和高通输出C;可再用B和C合成一个信号D。可逆变换中(理想地D=A),当然D或D与A的平均,都可以替换A,这种冗余太平庸无益了。不能确定有益的考虑是:以较小的耗时量等为代价,怎样识别某些重要特征,分别对A、B、C都做其它(例如降噪的软阈值)处理,且确定和利用然后D和A的关系。
图片1. 给实变换中滤波器序列添加噪声化相谱的例子和结果
图片2. 离散傅立叶变换、CQF和qmf的时间零点和符号问题
附.Wikipedia中最新“Wavelet”摘录
https://en.wikipedia.org/wiki/Wavelet
There are a large number of wavelet transforms each suitable for different applications. For a full list see list of wavelet
-related transforms but the common ones are listed below:
Continuous wavelet transform (CWT)
Discrete wavelet transform (DWT)
Fast wavelet transform (FWT)
Lifting scheme & Generalized Lifting Scheme
Wavelet packet decomposition (WPD)
Stationary wavelet transform (SWT)
Fractional Fourier transform (FRFT)
Fractional wavelet transform (FRWT)
Generalized transforms
There are a number of generalized transforms of which the wavelet transform is a special case.
For example, Joseph Segman introduced scale into the Heisenberg group, giving rise to a continuous transform space that is a
function of time, scale, and frequency. The CWT is a two-dimensional slice through the resulting 3d time-scale-frequency
volume.
Another example of a generalized transform is the chirplet transform in which the CWT is also a two dimensional slice through
the chirplet transform.
An important application area for generalized transforms involves systems in which high frequency resolution is crucial. For
example, darkfield electron optical transforms intermediate between direct and reciprocal space have been widely used in the
harmonic analysis of atom clustering, i.e. in the study of crystals and crystal defects. Now that transmission electron
microscopes are capable of providing digital images with picometer-scale information on atomic periodicity in nanostructure
of all sorts, the range of pattern recognition and strain/metrology applications for intermediate transforms with
high frequency resolution (like brushlets and ridgelets) is growing rapidly.
Fractional wavelet transform (FRWT) is a generalization of the classical wavelet transform in the fractional Fourier
transform domains. This transform is capable of providing the time- and fractional-domain information simultaneously and
representing signals in the time-fractional-frequency plane.
List of wavelets
Discrete wavelets
Beylkin (18)
BNC wavelets
Coiflet (6, 12, 18, 24, 30)
Cohen-Daubechies-Feauveau wavelet (Sometimes referred to as CDF N/P or Daubechies biorthogonal wavelets)
Daubechies wavelet (2, 4, 6, 8, 10, 12, 14, 16, 18, 20, etc.)
Binomial-QMF (Also referred to as Daubechies wavelet)
Haar wavelet
Mathieu wavelet
Legendre wavelet
Villasenor wavelet
Symlet
Continuous wavelets
Real-valued
Beta wavelet
Hermitian wavelet
Hermitian hat wavelet
Meyer wavelet
Mexican hat wavelet
Poisson wavelet
Shannon wavelet
Spline wavelet
Stromberg wavelet
Complex-valuedComplex Mexican hat wavelet
fbsp wavelet
Morlet wavelet
Shannon wavelet
Modified Morlet wavelet
This page was last modified on 2 December 2015, at 15:16.
转自:http://blog.sina.com.cn/s/blog_729a92140102w3ci.html
页:
[1]