Python numpy.kaiser方法代碼示例

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def smooth_curve(x):
    """用於使損失函數的圖形變圓滑
    參考：http://glowingpython.blogspot.jp/2012/02/convolution-with-numpy.html
    """
    window_len = 11
    s = np.r_[x[window_len-1:0:-1], x, x[-1:-window_len:-1]]
    w = np.kaiser(window_len, 2)
    y = np.convolve(w/w.sum(), s, mode='valid')
    return y[5:len(y)-5] 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def window_visu(N=51, name='hamming', **kargs):
    """A Window visualisation tool

    :param N: length of the window
    :param name: name of the window
    :param NFFT: padding used by the FFT
    :param mindB: the minimum frequency power in dB
    :param maxdB: the maximum frequency power in dB
    :param kargs: optional arguments passed to :func:`create_window`

    This function plot the window shape and its equivalent in the Fourier domain.

    .. plot::
        :width: 80%
        :include-source:

        from spectrum import window_visu
        window_visu(64, 'kaiser', beta=8.)

    """
    # get the default parameters
    mindB = kargs.pop('mindB', -100)
    maxdB = kargs.pop('maxdB', None)
    norm = kargs.pop('norm', True)

    # create a window object
    w = Window(N, name, **kargs)

    # plot the time and frequency windows
    w.plot_time_freq(mindB=mindB, maxdB=maxdB, norm=norm) 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def smooth_curve(x):
    """損失関數のグラフを滑らかにするために用いる

    參考：http://glowingpython.blogspot.jp/2012/02/convolution-with-numpy.html
    """
    window_len = 11
    s = np.r_[x[window_len-1:0:-1], x, x[-1:-window_len:-1]]
    w = np.kaiser(window_len, 2)
    y = np.convolve(w/w.sum(), s, mode='valid')
    return y[5:len(y)-5] 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def kaiser(M, beta):
    """Return the Kaiser window.
    The Kaiser window is a taper formed by using a Bessel function.

    .. math::  w(n) = I_0\\left( \\beta \\sqrt{1-\\frac{4n^2}{(M-1)^2}}
               \\right)/I_0(\\beta)

    with

    .. math:: \\quad -\\frac{M-1}{2} \\leq n \\leq \\frac{M-1}{2}

    where :math:`I_0` is the modified zeroth-order Bessel function.

     Args:
        M (int):
            Number of points in the output window. If zero or less, an empty
            array is returned.
        beta (float):
            Shape parameter for window

    Returns:
        ~cupy.ndarray:  The window, with the maximum value normalized to one
        (the value one appears only if the number of samples is odd).

    .. seealso:: :func:`numpy.kaiser`
    """
    if M == 1:
        return cupy.array([1.])
    if M <= 0:
        return cupy.array([])
    alpha = (M - 1) / 2.0
    out = cupy.empty(M, dtype=cupy.float64)
    return _kaiser_kernel(beta, alpha, out) 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def _kaiser(n, beta):
    """Independant Kaiser window

    For the definition of the Kaiser window, see A. V. Oppenheim & R. W. Schafer, "Discrete-Time Signal Processing".

    The continuous version of width n centered about x=0 is:

    .. note:: 2 times slower than scipy.kaiser
    """
    from scipy.special import iv as besselI
    m = n - 1
    k = arange(0, m)
    k = 2. * beta / m * sqrt (k * (m - k))
    w = besselI (0, k) / besselI (0, beta)
    return w 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def iFFT(Y, output_length=None, window=False):
    """ Inverse real-valued Fourier Transform

    Parameters
    ----------
    Y : array_like
       Frequency domain data [Nsignals x Nbins]
    output_length : int, optional
       Length of returned time-domain signal (Default: 2 x len(Y) + 1)
    window : boolean, optional
       Window applied to the resulting time-domain signal

    Returns
    -------
    y : array_like
       Reconstructed time-domain signal
    """
    Y = _np.atleast_2d(Y)
    y = _np.fft.irfft(Y, n=output_length)

    if window:
        no_of_samples = y.shape[-1]

        if window == 'hann':
            window_array = _np.hanning(no_of_samples)
        elif window == 'hamming':
            window_array = _np.hamming(no_of_samples)
        elif window == 'blackman':
            window_array = _np.blackman(no_of_samples)
        elif window == 'kaiser':
            window_array = _np.kaiser(no_of_samples, 3)
        else:
            raise ValueError('Selected window must be one of hann, hamming, blackman or kaiser')

        y *= window_array

    return y


# noinspection PyUnusedLocal 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def smooth_curve(x):
    """用於使損失函數的圖形變圓滑

    參考：http://glowingpython.blogspot.jp/2012/02/convolution-with-numpy.html
    """
    window_len = 11
    s = np.r_[x[window_len-1:0:-1], x, x[-1:-window_len:-1]]
    w = np.kaiser(window_len, 2)
    y = np.convolve(w/w.sum(), s, mode='valid')
    return y[5:len(y)-5] 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def _resample_window_oct(p, q):
    """Port of Octave code to Python"""

    gcd = np.gcd(p, q)
    if gcd > 1:
        p /= gcd
        q /= gcd

    # Properties of the antialiasing filter
    log10_rejection = -3.0
    stopband_cutoff_f = 1. / (2 * max(p, q))
    roll_off_width = stopband_cutoff_f / 10

    # Determine filter length
    rejection_dB = -20 * log10_rejection
    L = np.ceil((rejection_dB - 8) / (28.714 * roll_off_width))

    # Ideal sinc filter
    t = np.arange(-L, L + 1)
    ideal_filter = 2 * p * stopband_cutoff_f \
        * np.sinc(2 * stopband_cutoff_f * t)

    # Determine parameter of Kaiser window
    if (rejection_dB >= 21) and (rejection_dB <= 50):
        beta = 0.5842 * (rejection_dB - 21)**0.4 \
            + 0.07886 * (rejection_dB - 21)
    elif rejection_dB > 50:
        beta = 0.1102 * (rejection_dB - 8.7)
    else:
        beta = 0.0

    # Apodize ideal filter response
    h = np.kaiser(2 * L + 1, beta) * ideal_filter

    return h 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def smooth_curve(x):
    """ Used to smooth the graph of the loss function

    reference:http://glowingpython.blogspot.jp/2012/02/convolution-with-numpy.html
    """
    window_len = 11
    s = np.r_[x[window_len-1:0:-1], x, x[-1:-window_len:-1]]
    w = np.kaiser(window_len, 2)
    y = np.convolve(w/w.sum(), s, mode='valid')
    return y[5:len(y)-5] 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def kaiser(active, *, beta):
    """Kaiser tapering window.

    This uses :func:`numpy.kaiser`.

    Parameters
    ----------
    active : array_like, dtype=bool
        A boolean array containing ``True`` for active loudspeakers.
    alpha : float
        Shape parameter of the Kaiser window, see :func:`numpy.kaiser`.

    Returns
    -------
    (len(active),) `numpy.ndarray`
        Tapering weights.

    Examples
    --------
    .. plot::
        :context: close-figs

        plt.plot(sfs.tapering.kaiser(active1, beta=0), label='beta = 0')
        plt.plot(sfs.tapering.kaiser(active1, beta=2), label='beta = 2')
        plt.plot(sfs.tapering.kaiser(active1, beta=6), label='beta = 6')
        plt.plot(sfs.tapering.kaiser(active1, beta=8.6), label='beta = 8.6')
        plt.plot(sfs.tapering.kaiser(active1, beta=14), label='beta = 14')
        plt.axis([-3, 103, -0.1, 1.1])
        plt.legend(loc='lower center')

    .. plot::
        :context: close-figs

        plt.plot(sfs.tapering.kaiser(active2, beta=7))
        plt.axis([-3, 103, -0.1, 1.1])

    """
    idx = _windowidx(active)
    window = _np.zeros(len(active))
    window[idx] = _np.kaiser(len(idx), beta)
    return window 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def window_design(self, window_length, beta):
        """Kaiser window design

        Args:
            window_length: Length of the window in number of samples
            beta: Beta value for Kaiser window design

        Returns:
            window: Window designed using the beta and length provided as inputs

        """

        self.window = np.kaiser(window_length, beta)

        return self.window 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def enbw(data):
    r"""Computes the equivalent noise bandwidth

    .. math:: ENBW = N \frac{\sum_{n=1}^{N} w_n^2}{\left(\sum_{n=1}^{N} w_n \right)^2}

    .. doctest::

        >>> from spectrum import create_window, enbw
        >>> w = create_window(64, 'rectangular')
        >>> enbw(w)
        1.0

    The following table contains the ENBW values for some of the
    implemented windows in this module (with N=16384). They have been
    double checked against litterature (Source: [Harris]_, [Marple]_).

    If not present, it means that it has not been checked.

    =================== ============ =============
    name                 ENBW        litterature
    =================== ============ =============
    rectangular         1.           1.
    triangle            1.3334       1.33
    Hann                1.5001       1.5
    Hamming             1.3629       1.36
    blackman            1.7268       1.73
    kaiser              1.7
    blackmanharris,4    2.004        2.
    riesz               1.2000       1.2
    riemann             1.32         1.3
    parzen              1.917        1.92
    tukey 0.25          1.102        1.1
    bohman              1.7858       1.79
    poisson 2           1.3130       1.3
    hanningpoisson 0.5  1.609        1.61
    cauchy              1.489        1.48
    lanczos             1.3
    =================== ============ =============


    """
    N = len(data)
    return N * np.sum(data**2) / np.sum(data)**2 

# 需要導入模塊: import numpy [as 別名]
# 或者: from numpy import kaiser [as 別名]
def define_weight_function(self, weight_size=DEFAULT_WEIGHT_SIZE):
        """
        Try to derive WgtFunct from WgtType, if necessary. This should likely be called from the `GridType` parent.

        Parameters
        ----------
        weight_size : int
            the size of the `WgtFunct` to generate.

        Returns
        -------
        None
        """

        if self.WgtType is None or self.WgtType.WindowName is None:
            return  # nothing to be done

        window_name = self.WgtType.WindowName.upper()
        if window_name == 'HAMMING':
            # A Hamming window is defined in many places as a raised cosine of weight .54, so this is the default.
            # Some data use a generalized raised cosine and call it HAMMING, so we allow for both uses.
            try:
                # noinspection PyTypeChecker
                coef = float(self.WgtType.get_parameter_value(None, 0.54))  # just get first parameter - name?
            except ValueError:
                coef = 0.54
            self.WgtFunct = _raised_cos(weight_size, coef)
        elif window_name == 'HANNING':
            self.WgtFunct = _raised_cos(weight_size, 0.5)
        elif window_name == 'KAISER':
            try:
                # noinspection PyTypeChecker
                beta = float(self.WgtType.get_parameter_value(None, 14))  # just get first parameter - name?
            except ValueError:
                beta = 14.0  # default suggested in numpy.kaiser
            self.WgtFunct = numpy.kaiser(weight_size, beta)
        elif window_name == 'TAYLOR':
            # noinspection PyTypeChecker
            sidelobes = int(self.WgtType.get_parameter_value('NBAR', 4))  # apparently the matlab argument name
            # noinspection PyTypeChecker
            max_sidelobe_level = float(self.WgtType.get_parameter_value('SLL', -30))  # same
            if max_sidelobe_level > 0:
                max_sidelobe_level *= -1
            self.WgtFunct = _taylor_win(weight_size,
                                        sidelobes=sidelobes,
                                        max_sidelobe_level=max_sidelobe_level,
                                        normalize=True)