def wpi(dataset, n_channels, seq_length):
"""
function to calculate a wavelet packet energy image
# Ding.2017: Energy-Fluctuated Multiscale Feature Learning With Deep ConvNet for Intelligent Spindle Bearing Fault Diagnosis
:param dataset: the raw data read from .csv or .npy file
:type dataset: ndarray
:param n_channels: number of channels
:type n_channels: int
:param seq_length: length of one data stream of a single sensor
:type seq_length: int
:return: the flattened images, tupel holding the image size
"""
level = 10 # choose an even number, paper uses 10
wavelet = 'db8' # Daubechies 8 used in paper
# wavelet = 'coif3' # Daubechies 8 used in paper
order = "natural" # other option is "freq"
clip_energy = 1 # threshold to clip the calculated energy features (negative and positive)
num_samples = dataset.shape[0]
img_size = np.power(2, level // 2)
size_flat = img_size * img_size
pics = np.zeros([num_samples, n_channels * size_flat]) # initialize array
wp_image = np.zeros([img_size, img_size])
for sample in range(num_samples): # loop over all samples
for ch in range(n_channels): # loop over all channels
# Construct wavelet packet tree for signal of one channel
wp = WaveletPacket(dataset[sample][ch * seq_length:(ch + 1) *
seq_length],
wavelet,
'symmetric',
maxlevel=level)
nodes = wp.get_level(
level, order=order
) # !required! access the tree to populate it (might be a bug of the library?)
i = 0
# loop through the tree (ordered from aaa..a to ddd..d)
for node in wp.get_leaf_nodes():
# use only the coefficients from node (i, p) with p = 0..(2^i-1), i.e. set all other coefficients to zero
new_wp = WaveletPacket(None, wavelet, 'symmetric', level)
new_wp[node.path] = wp[node.path].data
# get the reconstruction coefficients (length 2^i)
reconst = np.asarray(new_wp.reconstruct(update=False))
# phase shift --> arrange energy features calculated as the squared sum over the reconstruction coefficients
wp_image[i % img_size,
i // img_size] = np.sum(np.multiply(reconst, reconst),
axis=-1)
# remove node from wp tree
del new_wp[node.path]
i += 1
# (!) THP modification (!), clip the wpi to fixed range: especially the approximation coefficients hold a lot of energy which
# scales very differently
wp_image = np.clip(wp_image, -clip_energy, clip_energy)
# collect all pictures, shape (samples, ch1+ch2..)
pics[sample][ch * size_flat:(ch + 1) * size_flat] = np.reshape(
wp_image, [1, size_flat])
return pics, [img_size, img_size]
def wavlet_barkms(x,wname,fs):
"""
decompose speech into 5 layers wavelet packet according to wnmae
:param x:
:param wname: wavelet generating function
:param fs: 8000
:return y: 17 BASK sub-band filter
"""
if fs != 8000:
print('fs must be 8000Hz, change fs!!!')
return
y = np.zeros((17, len(x)))
n = 5 # decomposition level
T = WaveletPacket(data=x, wavelet=wname, maxlevel=n, mode='zero') # 一维小波包分解
new_wp = WaveletPacket(data=np.zeros(len(x)), wavelet='db2', mode='zero')
# 计算各节点对应系数
map = []
NodeName = []
for row in range(n):
map.append([])
NodeName.append([])
for i in [node.path for node in T.get_level(level=row + 1, order='natural')]:
map[row].append(T[i].data)
NodeName[row].append(i)
# 按指定的节点，对时间序列分解的一位小波包系数重构
for i in range(8):
new_wp[NodeName[4][i]] = map[4][i]
y[i, :] = new_wp.reconstruct(update=False)
new_wp = WaveletPacket(data=np.zeros(len(x)), wavelet=wname, mode='zero')
new_wp[NodeName[3][4]] = map[3][4]
y[8, :] = new_wp.reconstruct(update=False)
new_wp = WaveletPacket(data=np.zeros(len(x)), wavelet=wname, mode='zero')
new_wp[NodeName[3][5]] = map[3][5]
y[9, :] = new_wp.reconstruct(update=False)
new_wp = WaveletPacket(data=np.zeros(len(x)), wavelet=wname, mode='zero')
new_wp[NodeName[4][11]] = map[4][11]
y[10, :] = new_wp.reconstruct(update=False)
new_wp = WaveletPacket(data=np.zeros(len(x)), wavelet=wname, mode='zero')
new_wp[NodeName[4][12]] = map[4][12]
y[11, :] = new_wp.reconstruct(update=False)
new_wp = WaveletPacket(data=np.zeros(len(x)), wavelet=wname, mode='zero')
new_wp[NodeName[3][7]] = map[3][7]
y[12, :] = new_wp.reconstruct(update=False)
new_wp = WaveletPacket(data=np.zeros(len(x)), wavelet=wname, mode='zero')
for i in range(4, 8):
new_wp[NodeName[2][i]] = map[2][i]
y[9 + i, :] = new_wp.reconstruct(update=False)
new_wp = WaveletPacket(data=np.zeros(len(x)), wavelet=wname, mode='zero')
return y
def getSumOfBandsPercentages(signal):
wp = WaveletPacket(data = signal, wavelet = 'db4', maxlevel = 6)
nodes = [node.path for node in wp.get_level(6, 'natural')]
delta = np.sum(np.abs(wp[nodes[0]].data))
theta = np.sum(np.abs(wp[nodes[1]].data))
alpha = np.sum(np.abs(wp[nodes[2]].data))
beta = 0
gamma = 0
for i in range(3, 9):
beta += np.sum(np.abs(wp[nodes[i]].data))
for i in range(9, 14):
gamma += np.sum(np.abs(wp[nodes[i]].data))
bands = [delta, theta, alpha, beta, gamma]
sum = np.sum(bands)
percentages = bands/sum
percentages *= 100
return percentages
def apply_wavelet_ICA(EEGdata):
""" apply wavelet decomposition and ICA to EEGdata """
terminal_nodes = ['aa', 'ad', 'da',
'dd'] # labels for level 2 nodes in WaveletPacket tree
T = len(terminal_nodes)
N, _ = EEGdata.shape # no. channels
wavePacks = []
waveData = [[] for i in range(T)]
for c in range(N):
wavePacket = WaveletPacket(data=EEGdata[c, :],
wavelet='sym4',
maxlevel=2)
wavePacks.append(wavePacket)
for i, n in enumerate(terminal_nodes):
waveData[i].append(wavePacket[n].data)
mixMat, V, X = sobi_fast(np.array(waveData[0]))
ICs = [np.matmul(np.transpose(V), X)]
ICs += [np.array(wp) for wp in waveData[1:]]
return ICs, mixMat, wavePacks, terminal_nodes
def wave_fea(a):
wp = WaveletPacket(a, 'db1', maxlevel=8)
nodes = wp.get_level(8, "freq")
return np.linalg.norm(np.array([n.data for n in nodes]), 2)
import pandas as pd
from pywt import WaveletPacket
import pywt.data
ecg = pywt.data.ecg()
df = pd.read_csv('sp.csv')
tmp = df['Close'].tolist()
tmp2 = df['Open'].tolist()
diff = []
for i in range(0, len(tmp)):
diff.append(tmp[i] - tmp2[i])
data3 = diff
wp = WaveletPacket(data3, 'sym5', maxlevel=4)
fig = plt.figure()
plt.set_cmap('bone')
ax = fig.add_subplot(wp.maxlevel + 1, 1, 1)
ax.plot(ecg, 'k')
ax.set_xlim(0, len(ecg) - 1)
ax.set_title("Wavelet packet coefficients")
for level in range(1, wp.maxlevel + 1):
ax = fig.add_subplot(wp.maxlevel + 1, 1, level + 1)
nodes = wp.get_level(level, "freq")
nodes.reverse()
labels = [n.path for n in nodes]
values = -abs(np.array([n.data for n in nodes]))
ax.imshow(values, interpolation='nearest', aspect='auto')
#!/usr/bin/env python
# -*- coding: utf-8 -*-
import os
import numpy as np
import matplotlib.pyplot as plt
from pywt import WaveletPacket
import pywt.data
ecg = pywt.data.ecg()
wp = WaveletPacket(ecg, 'sym5', maxlevel=4)
fig = plt.figure()
plt.set_cmap('bone')
ax = fig.add_subplot(wp.maxlevel + 1, 1, 1)
ax.plot(ecg, 'k')
ax.set_xlim(0, len(ecg) - 1)
ax.set_title("Wavelet packet coefficients")
for level in range(1, wp.maxlevel + 1):
ax = fig.add_subplot(wp.maxlevel + 1, 1, level + 1)
nodes = wp.get_level(level, "freq")
nodes.reverse()
labels = [n.path for n in nodes]
values = -abs(np.array([n.data for n in nodes]))
ax.imshow(values, interpolation='nearest', aspect='auto')
ax.set_yticks(np.arange(len(labels) - 0.5, -0.5, -1), labels)
plt.setp(ax.get_xticklabels(), visible=False)
#!/usr/bin/env python
# -*- coding: utf-8 -*-
from pywt import WaveletPacket
wp = WaveletPacket(range(16), 'db2', maxlevel=3)
print[node.path for node in wp.get_leaf_nodes(decompose=False)]
print[node.path for node in wp.get_leaf_nodes(decompose=True)]
coeffs = [(node.path, node.data) for node in wp.get_leaf_nodes(decompose=True)]
print coeffs
wp2 = WaveletPacket(None, 'db2', maxlevel=3)
for path, data in coeffs:
wp2[path] = data
#print wp["a"]
print[node.path for node in wp2.get_leaf_nodes(decompose=False)]
print wp2.reconstruct()
