feature woof

feature_wood.py

def wavelet_analysis(sig):
    """ decompose signal into wavelets
        returns the frequence of the thesholded filtered signal
    """
    widths = np.arange(1, 100,1)
    cwtmatr = signal.cwt(sig, signal.ricker, widths)

    N = sig.size
    fsig = cwtmatr[98,:]
    detect_pos = fsig>30
    detect_neg = fsig<-30
    n_detect_pos = np.sum(np.diff(detect_pos))/2
    n_detect_neg = np.sum(np.diff(detect_neg))/2
    f_detect_pos = 1.*n_detect_pos / N
    f_detect_neg = 1.*n_detect_neg / N
    power_spectrum = 1.*np.sum(np.abs(fsig))/N
    return {'n_detect_pos':n_detect_pos, 'f_detect_pos': f_detect_pos,
            'n_detect_neg':n_detect_neg, 'f_detect_neg': f_detect_neg,
            'fsig': fsig,'N':N, 'power':power_spectrum}