MaNoVe.wst(wavethresh)
MaNoVe.wst()所属R语言包：wavethresh
Make Node Vector (using Coifman-Wickerhauser best-basis type algorithm) on nondecimated wavelet transform object
节点矢量（夸夫曼Wickerhauser最好的基础类型算法）nondecimated小波变换对象
译者：生物统计家园网 机器人LoveR
描述----------Description----------
This method chooses a "best-basis" using the Coifman-Wickerhauser (1992) algorithm applied to nondecimated wavelet transform, wst.object, objects.
这种方法选择“最基础”的使用的Coifman Wickerhauser（1992）nondecimated小波变换算法，wst.object“对象。
用法----------Usage----------
## S3 method for class 'wst':
MaNoVe(wst, entropy=Shannon.entropy, verbose=FALSE,
stopper=FALSE, alg="C", ...)
参数----------Arguments----------
参数：wst
The wst object for which you wish to find the best basis for.
WST对象，你想找到最好的基础。
参数：entropy
The function used for computing the entropy of a vector
的功能，用于计算一个矢量的熵
参数：verbose
Whether or not to print out informative messages
无论打印出的信息性消息
参数：stopper
Whether the computations are temporarily stopped after each packet. This can be useful in conjunction with the verbose argument so as to see computations proceed one packet at a time.
是否计算每个数据包后，暂时停止。一并verbose参数，以看到计算的时间继续进行一个分组，这可能是有用的。
参数：alg
If "C" then fast compiled C code is used (in which case the entropy function is ignored and the C code uses an internal Shannon entropy. Otherwise, slower R code is used but an arbitrary entropy argument can be used
如果“C”，然后快速编译的C代码（在这种情况下，被忽略entropy函数和C代码使用一个内部的香农熵，否则，速度较慢的R代码的任意entropy参数可以使用
参数：...
Other arguments
其他参数
Details
详细信息----------Details----------
Description says all
介绍说，所有的
值----------Value----------
A wavelet node vector object, of class nv, a basis description. This can be fed into a basis inversion using, say, the function InvBasis.
小波的的节点矢量对象，类nv“的基础描述。这可以被送入一个基础反转，也就是说，使用的功能InvBasis。
（作者）----------Author(s)----------
G P Nason
参见----------See Also----------
InvBasis, MaNoVe, MaNoVe.wp, Shannon.entropy, wst.object, wst
InvBasis，MaNoVe，MaNoVe.wp，Shannon.entropy，wst.object，wst
实例----------Examples----------
#[]
# What follows is a simulated denoising example. We first create our[下面是一个模拟去噪的例子。我们首先创建]
# "true" underlying signal, v. Then we add some noise to it with a signal[“真正”的相关信号，v。然后我们添加一些噪声信号]
# to noise ratio of 6. Then we take the packet-ordered non-decimated wavelet[噪声比为6。然后，我们把数据包排序，非抽取小波]
# transform and then threshold that.[改造和阈值。]
#[]
# Then, to illustrate this function, we compute a "best-basis" node vector[然后，来说明这个功能，我们计算“最基础”的节点矢量]
# and use that to invert the packet-ordered NDWT using this basis. As a[并使用反转分组排序NDWT此基础上。作为一个]
# comparison we also use the Average Basis method[比较，我们也可以使用平均法计算方法]
# (cf Coifman and Donoho, 1995). [（CF Coifman和Donoho提出，1995年）。]
#[]
# NOTE: It is IMPORTANT to note that this example DOES not necessarily[注：重要的是要注意，这个例子不一定]
# use an appropriate or good threshold or necessarily the right underlying[使用适当的还是不错的阈值，或者一定是正确的底层]
# wavelet. I am trying to show the general idea and please do not "quote" this[小波。我试图表明的总体思路和请不要“注明”]
# example in literature saying that this is the way that WaveThresh (or[例如，在文献说，这是方式，WaveThresh（或]
# any of the associated authors whose methods it attempts to implement)[任何相关的作者的方法，它试图实现）]
# does it. Proper denoising requires a lot of care and thought.[这样做了。正确的去噪需要大量的良苦用心。]
#[]
#[]
# Here we go....[在这里，我们离开了....]
#[]
# Create an example vector (the Donoho and Johnstone heavisine function)[创建一个示例矢量（在Donoho和Johnstone heavisine的功能）]
#[]
v <- DJ.EX()$heavi
#[]
# Add some noise with a SNR of 6[添加一些噪声，信噪比为6]
#[]
vnoise <- v + rnorm(length(v), 0, sd=sqrt(var(v))/6)
#[]
# Take packet-ordered non-decimated wavelet transform (note default wavelet[以分组下令非抽取小波变换（附注默认小波]
# used which might not be the best option for denoising performance).[这可能不是最好的选择，消噪性能）。]
#[]
vnwst <- wst(vnoise)
#[]
# Let's take a look at the wavelet coefficients of vnoise[让我们来看看在小波系数VNOISE]
#[]
## Not run: plot(vnwst)[＃不运行：图（vnwst）]
#[]
# Wow! A huge number of coefficients, but mostly all noise.[哇！数量庞大的系数，但大多都是噪音。]
#[]
#[]
# Threshold the resultant NDWT object.[阈值的的结果NDWT对象。]
# (Once again default arguments are used which are certainly not optimal).[（再次使用默认参数肯定不是最优的）。]
#[]
vnwstT <- threshold(vnwst)
#[]
# Let's have a look at the thresholded wavelet coefficients[让我们一起来看看阈值的小波系数]
#[]
## Not run: plot(vnwstT)[＃不运行：图（vnwstT）]
#[]
# Ok, a lot of the coefficients have been removed as one would expect with[好了，有很多的系数已被删除，正如人们所期望的]
# universal thresholding[普遍的阈值]
#[]
#[]
# Now select packets for a basis using a Coifman-Wickerhauser algorithm[现在选择的数据包使用夸夫曼Wickerhauser算法的基础]
#[]
vnnv <- MaNoVe(vnwstT)
#[]
# Let's have a look at which packets got selected[让我们看看哪些信息包是]
#[]
vnnv
# Level : 9 Action is R (getpacket Index: 1 )[等级：9行动是R（getpacket指数：1）]
# Level : 8 Action is L (getpacket Index: 2 )[等级：8行动的L（getpacket指数：2）]
# Level : 7 Action is L (getpacket Index: 4 )[等级：7行动是L（getpacket指数：4）]
# Level : 6 Action is L (getpacket Index: 8 )[等级：6行动是L（getpacket指数：8）]
# Level : 5 Action is R (getpacket Index: 17 )[等级：5行动是R（getpacket指数：17）]
# Level : 4 Action is L (getpacket Index: 34 )[等级：4行动是L（getpacket指数：34）]
# Level : 3 Action is L (getpacket Index: 68 )[等级：3行动是L（getpacket指数：68）]
# Level : 2 Action is R (getpacket Index: 137 )[等级：2动作是R（getpacket指数：137）]
# Level : 1 Action is R (getpacket Index: 275 )[等级：1动作是R（getpacket指数：275）]
# There are 10 reconstruction steps[有10个重建的步骤]
#[]
# So, its not the regular decimated wavelet transform![因此，它不是普通的抽取小波变换！]
#[]
# Let's invert the representation with respect to this basis defined by[让我们反转表示此基础上定义的]
# vnnv[vnnv]
#[]
vnwrIB <- InvBasis(vnwstT, vnnv)
#[]
# And also, for completeness let's do an Average Basis reconstruction.[同时，完整性，让我们做一个平均法计算重建。]
#[]
vnwrAB <- AvBasis(vnwstT)
#[]
# Let's look at the Integrated Squared Error in each case.[让我们来看看在每一种情况下的积分平方误差。]
#[]
sum( (v - vnwrIB)^2)
# [1] 386.2501[[1] 386.2501]
#[]
sum( (v - vnwrAB)^2)
# [1] 328.4520[[1] 328.4520]
#[]
# So, for this limited example the average basis method does better. Of course,[因此，对于这有限的例子中，平均基准计算方法没有更好的。当然，]
# for *your* simulation it could be the other way round. "Occasionally", the[*你*模拟的，也可能是反过来。 “偶尔”]
# inverse basis method does better. When does this happen? A good question.[逆基础方法没有更好的。当出现这种情况呢？一个很好的问题。]
#[]
# Let's plot the reconstructions and also the original[让我们绘制的重建，也是原]
#[]
## Not run: plot(vnwrIB, type="l")[＃不运行：的图（vnwrIB，类型为“L”）]
## Not run: lines(vnwrAB, lty=2)[＃不运行：行（vnwrAB，LTY = 2）]
## Not run: lines(v, lty=3)[＃不运行线（V，LTY = 3）]
#[]
# The dotted line is the original. Neither reconstruction picks up the[虚线是原始的。无论是重建拿起]
# spikes in heavisine very well. The average basis method does track the[在heavisine高峰很好。平均法计算的方法跟踪]
# original signal more closely though.[原始信号更加紧密了。]
#[]
转载请注明:出自 生物统计家园网(http://www.biostatistic.net)。
注：
注1：为了方便大家学习，本文档为生物统计家园网机器人LoveR翻译而成，仅供个人R语言学习参考使用，生物统计家园保留版权。
注2：由于是机器人自动翻译，难免有不准确之处，使用时仔细对照中、英文内容进行反复理解，可以帮助R语言的学习。
注3：如遇到不准确之处，请在本贴的后面进行回帖，我们会逐渐进行修订。
|
发表于 2012-10-1 17:26:28
