wavCWTPeaks(wmtsa)
wavCWTPeaks()所属R语言包：wmtsa
Peak detection in a time series via the CWT
通过CWT在一个时间序列的峰值检测
译者：生物统计家园网 机器人LoveR
描述----------Description----------
Finds the local maxima in a time series via a CWT tree.
查找CWT的树通过在时间序列中的局部最大值。
用法----------Usage----------
参数----------Arguments----------
参数：x
an object of class wavCWTTree.
对象类wavCWTTree。
参数：length.min
the minimum number of points along a CWT tree branch and within the specified scale.range needed in order for that branches peak to be considered a peak candidate. Default: 10.
CWT树分支，并在指定的scale.range的最低点沿所需要为了该部门峰值被认为是高峰的候选人。默认值：10。
参数：noise.fun
a character string defining the function to apply to the local noise estimates in order to sumarize and quantify the local noise level into a scalar value. See the DETAILS section for more information. Supported values are
一个字符串定义的功能，适用于当地的噪音估计，以sumarize当地的噪音水平和量化成一个标值。更多信息，请参阅详细信息部分。支持的值
"quantile"quantile(x, probs=0.95)
“位数”位数（X，probs = 0.95）
"sd"sd(x)
“SD”SD（X）
"mad"mad(x, center=0)
“疯狂”狂（X，中心= 0）
where x is a vector of smallest-scale CWT coefficients whose time indices are near that of the branch termination time. Default: "quantile".
x是一个向量的最小规模的CWT系数，其时间指数是附近的分公司终止时间。默认值："quantile"。
参数：noise.min
the minimum allowed estimated local noise level. Default: quantile(attr(x,"noise"), prob=0.05), where x is the input wavCWTTree object.
允许的最小估计当地的噪音水平。默认值：quantile(attr(x,"noise"), prob=0.05)，其中x是输入wavCWTTree对象。
参数：noise.span
the span in time surrounding each branche's temrination point to use in forming local noise estimates and (ultimately) peak SNR estimates. Default: NULL,max(0.01 * diff(range(times)), 5*sampling.interval), where times and sampling.interval are attributes of the input wavCWTTree object.
在时间的跨度，周围每个BRANCHE temrination点形成噪音估计并（最终）峰值信噪比估计。默认值：NULL,max(0.01 * diff(range(times)), 5*sampling.interval)，times和sampling.interval的输入wavCWTTree对象的属性。
参数：scale.range
the range of CWT scales that a peak must fall into in order to be considered a peak candidate. Default: scale[range(which(branch.hist > quantile(branch.hist,prob=0.8)))], where branch.hist is an attribute of the input wavCWTTree object. This default results in isolating the bulk of energetic CWT branches, but the user in encouraged to reduce the scale range in order to attenuate the computational burden of the peak detection scheme.
CWT尺度范围内的峰值下降到在被视为一个高峰候选人。默认值：scale[range(which(branch.hist > quantile(branch.hist,prob=0.8)))]，branch.hist输入wavCWTTree对象的属性。这个预设的结果，在的大部分精力充沛CWT分支隔离，但用户在鼓励减少的规模范围内的峰值检测方案，以减轻计算负担。
参数：snr.min
the minimum allowed peak signal-to-noise ratio. Default: 3.
允许的最小峰值信号噪声比。默认值：3。
Details
详细信息----------Details----------
The local maxima of the CWT are linked together to form so-called branches, where each branch represents one ridge of the CWT time-scale terrain. The collection of branches forms a tree, as output by the wavCWTTree function. The wavCWTpeaks function prunes the branches of the input CWT tree and records the termination time (i.e., the time associated with point of the branch that is closest to scale zero) as the time index associated with the local peak of the corresponding time series. Information regarding the collection of isolated peaks is returned as a data.frame object.
CWT的局部最大值的连接在一起，形成所谓的分支，每个分支表示一个脊的CWT的时间尺度地形。分支的集合形成了一个树，输出wavCWTTree功能。 wavCWTpeaks函数修剪输入CWT的树和记录终止时间（即，与点最接近缩放零分支相关联的时间）与相应的局部峰值作为相关联的时间索引的分支的时间序列。信息收集孤立的山峰作为一个data.frame对象返回。
The tree branches are pruned in the following ways:
被修剪树枝在以下方面：
peak SNRan estimate of SNR at peak value is greater than or equal to the specified snr.min. A peak SNR estimate is formed as follows: For each branch of the input CWT tree, a subset of CWT coefficients is collected such that the CWT coefficients are both local to the branch termination time and correspond to the smallest analyzed CWT scale. The user specified noise.span argument is used to define the boundaries of each subset in time ala [B - noise.span, B + noise.span], where B is the branch termination time. Each CWT subset is assumed to be representative of the local noise levels near the corresponding branch termination time and noise.fun is used to quantify (and summarize) each level resulting in a scalar $z$. The minimum value of $z$ is specified by the user ala the noise.min argument. Finally, the ratio |P|/|W| is used to form an estimate of the local signal-to-noise ration (SNR) for the corresponding branch, where P is the maximum CWT value along the branch in the CWT time-scale plane.
峰值SNRan峰值估计的信噪比（SNR）是大于或等于指定的snr.min。 A峰值SNR估计的形成如下：对于输入CWT树的每个分支，CWT系数的一个子集被收集，使得在CWT系数都是本地分支终止时间的对应的最小的CWT规模分析。用户指定了noise.span参数是用来定义每个子集的边界的时间阿拉[B - “noise.span，B + ”noise.span“]，其中 B的分公司终止时间。每个CWT子集被认为是代表当地附近的噪音等级相应的分支终止时间和noise.fun使用量化（总结）每个级别产生一个标量$ Z $。 $ Z $的用户阿拉noise.min的参数指定的最小值。最后，比|P|/|W|被用来形成的本地信号 - 噪声比（SNR）的估计值的相应的分支，其中P是最大CWT的值沿着分支中CWT的时间大型平面。
scalethe scale corresponding to the peak is larger than the minimum of the specified scale.range.
scalethe规模对应于峰值的最小值大于指定的scale.range。
branch lengththe length of the branch within the specified scale.range is greater than or equal to the specified minimum length.min.
的分支的分支lengththe长度内的指定的scale.range是大于或等于规定的最小length.min。
endpointthe index of the terminating time of the branch is on the interval (W, N-W), where N is the length of the series and W is integer equivalent of 1/4 the length of the noise.span or 3, whichever is greater.
endpointthe索引的分支终止时间的时间间隔上(W, N-W)，其中N是该系列的长度和W是1/4的长度的整数，等于noise.span或3，取更大。
NOTE: For peak detection, the wavelet filters used to form the CWT must maintain an (approximate) zero phase property so that the CWT coefficients can be meaningfully aligned with the events of the original time series. Currently, only the so-called Mexican hat wavelet maintains this property due to the even-symmetry of the filter's impulse response. Therefore, only the Mexican hat wavelet ("gaussian2") is currently supported for CWT-based peak detection. See the wavCWTFilters and wavCWT function for more information.
注：峰值检测，形成CWT的小波滤波器必须保持（大约）零相位特性，因此，的CWT系数可以进行有意义的一致与原始时间序列的事件。目前，只有所谓的Mexican hat小波保持此属性由于偶数对称滤波器的脉冲响应。因此，只有在Mexican hat小波（“gaussian2”）目前支持CWT基于峰值检测。请参阅wavCWTFilters和wavCWT函数的更多信息，。
值----------Value----------
a list of x and y vectors identifying the peaks in the original time series. The pruning criteria (snr.min, scale.range, length.min, noise.span, noise.fun, noise.min) are attached are attached as attributes. In addition, a peaks attribute is attached and corresponds to a data.frame containing the following information for each peak:
列表x和y确定在原来的时间序列的峰的向量。修剪标准（snr.min，scale.range，length.min，noise.span，noise.fun，noise.min）被安装附加属性。此外，附加一个peaks属性和对应于一个data.frame含有每个峰的以下信息：
<table summary="R valueblock"> <tr valign="top"><td>branch</td> <td> index of the associated branch in the CWT tree</td></tr>
<table summary="R valueblock"> <tr valign="top"> <TD>branch </ TD> <TD>指数相关联的分支在CWT树</ TD> </ TR>
<tr valign="top"><td>itime</td> <td> index location in time</td></tr>
<tr valign="top"> <TD> itime </ TD> <TD>索引位置</ TD> </ TR>
<tr valign="top"><td>iscale</td> <td> index location in scale</td></tr>
<tr valign="top"> <TD> iscale </ TD> <TD>索引位置，规模</ TD> </ TR>
<tr valign="top"><td>time</td> <td> location in time</td></tr>
<tr valign="top"> <TD> time</ TD> <TD>时间位置</ TD> </ TR>
<tr valign="top"><td>scale</td> <td> location in scale</td></tr>
<tr valign="top"> <TD> scale</ TD> <TD>的位置，规模</ TD> </ TR>
<tr valign="top"><td>extrema</td> <td> CWT value</td></tr>
<tr valign="top"> <TD> extrema </ TD> <TD> CWT值</ TD> </ TR>
<tr valign="top"><td>iendtime</td> <td> index location of branch termination time, i.e., the index of the point in the time series corresponding to the current peak</td></tr> </table>
<tr valign="top"> <TD>iendtime </ TD> <TD>的分公司终止时间的索引位置，也就是说，指数的时间序列中的相应的峰值电流</ TD> </ TR> </ TABLE>
参考文献----------References----------
Pan Du, Warren A. Kibbe, and Simon M. Lin, “Improved peak detection in mass spectrum by incorporating continuous wavelet transform-based pattern matching", Bioinformatics, 22, 2059–2065 (2006).
J.F. Muzy, E. Bacry, and A. Arneodo., “The multifractal formalism revisited with wavelets.", International Journal of Bifurcation and Chaos, 4, 245–302 (1994).
参见----------See Also----------
实例----------Examples----------
## create linchirp series [＃创建linchirp系列]
linchirp <- make.signal("linchirp")
## calculate the CWT [＃计算CWT]
W <- wavCWT(linchirp)
## form CWT tree [＃形成CWT树]
z <- wavCWTTree(W)
## estimate the peak locations using default [＃估算峰值使用默认的位置]
## scale.range [＃scale.range]
p <- wavCWTPeaks(z)
## plot an overlay of the original series and the [＃图覆盖在原始的系列和]
## peaks [＃峰]
x <- as(linchirp@positions,"numeric")
y <- linchirp@data
plot(x, y, type="l", xlab="time", ylab="linchirp")
points(p, pch=16, col="red", cex=1.2)
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 22:52:52
