def task_mad(data):
"""
http://statsmodels.sourceforge.net/devel/generated/statsmodels.robust.scale.stand_mad.html
"""
mad = stand_mad(data)
return mad
def wavelet_transform(noisySignal):
'''
Method that applies wavelt tarnsform (Daubechies, decomposition level=12, filter size=12) to a given signal.
Inputs:
- noisySignal: noisy signal to be transformed (array [m])
Output: (signal) wavelet transform of noisySignal (array [m])
'''
########################################
from statsmodels.robust import stand_mad
########################################
levels = 12
denoised = pywt.wavedec(noisySignal, 'db12', level=levels, mode='per')
sigma = stand_mad(denoised[-1])
threshold = sigma * np.sqrt(2 * np.log(len(noisySignal)))
denoised = map(lambda x: pywt.thresholding.soft(x, value=threshold),
denoised)
signal = pywt.waverec(denoised, 'db12', mode='per')
return signal
def visualize(data, wavelet, sigma=None):
coefs = pywt.wavedec(data, wavelet)
if sigma is None:
sigma = stand_mad(coefs[-1])
thresh = sigma*np.sqrt(2*np.log(len(data)))
denoised = coefs[:]
denoised[1:] = (pywt.thresholding.soft(i, value=thresh) for i in
denoised[1:])
rec = pywt.waverec(denoised, wavelet)
plt.plot(data, 'r')
plt.plot(rec, 'g')
plt.show()
return rec
def apply_filter(self, array):
if len(array.shape)!=1:
return False
mlv = pywt.dwt_max_level(array.shape[0],pywt.Wavelet(self.type))
coeffs = pywt.wavedec(array.data, self.type, level=mlv, mode='per')
if self.auto_threshold:
sigma = stand_mad(coeffs[-1])
uthresh = sigma*np.sqrt(2*np.log(len(coeffs)))
self.threshold = uthresh
else:
uthresh = self.threshold
denoised = coeffs[:]
if self.thresholding==0: # Hard thresholding
denoised[1:] = (pywt.thresholding.hard(i, value=uthresh) for i in denoised[1:])
elif self.thresholding==1: # Soft thresholding
denoised[1:] = (pywt.thresholding.soft(i, value=uthresh) for i in denoised[1:])
signal = pywt.waverec(denoised, self.type, mode='per')
array.data = signal[0:array.shape[0]]
return True
parnh[xi][yi] = minimize(
res1,
par2hned.params,
args=(velx, spec, []),
)
resc[:, xi, yi] = parnh[xi][yi].residual
import pywt
from statsmodels.robust import stand_mad
wings = zeros(cube.shape)
for (x, y), val in ndenumerate(cube[0]):
spectra2 = cube[:, x, y]
noisy_coefs = pywt.wavedec(spectra2, 'db4', level=11, mode='per')
sigma = stand_mad(noisy_coefs[-1])
uthresh = sigma * np.sqrt(2 * np.log(len(spectra2)))
denoised = noisy_coefs[:]
denoised[:] = (pywt.thresholding.soft(i, value=uthresh)
for i in denoised[:])
signal = pywt.waverec(denoised, 'db4', mode='per')
wings[:, x, y] = signal[:-1]
def getImages(par):
dv1 = ma.zeros((nh3bg.shape[0], nh3bg.shape[1]))
sigma1 = ma.zeros((nh3bg.shape[0], nh3bg.shape[1]))
I1 = ma.zeros((nh3bg.shape[0], nh3bg.shape[1]))
def removeBaselinesWave(cube):
from statsmodels.robust import stand_mad
from scipy import stats
import numpy as np
import pywt
def plfunc(w0, I, sigma, k, b, x):
d = -0.5 * (((x - w0) / sigma)**2)
aa = np.exp(d) * I + k * (x - w0) + b
return aa
constants = np.array([0])
for (x, y), val in np.ndenumerate(cube[1, :, :]):
print(x, y)
spectra = cube[:, x, y]
frist = spectra - baseline(spectra, fitfunc, 600)
sigma = np.std(frist)
arrs = []
ca = []
pos = np.where(abs(frist) > 1.5 * sigma)[0]
for i in range(len(pos) - 1):
ca.append(pos[i])
if pos[i + 1] - pos[i] != 1:
if len(ca) > 1:
arrs.append(ca)
ca = []
arrs = np.array(arrs)
ids = []
for i in range(len(arrs)):
if len(arrs[i]) < 3:
ids.append(i)
arrs = np.delete(arrs, ids)
w = []
s = []
I = []
b = []
k = []
for ran in arrs:
f = round(np.array(ran).mean())
if (abs(frist[f]) >
1.5 * sigma) and f > 200 and f < len(spectra) - 200:
params = Parameters()
params.add('w0', f, min=f - 10, max=f + 10)
params.add('I', spectra.max(), min=-10, max=10)
params.add('sigma', 1, min=0.01, max=10)
params.add('k', 0)
params.add('b', 1)
xr = np.arange(f - 100, f + 100, 1, dtype=int)
minn = minimize(resgaus,
params,
args=(xr, spectra[f - 100:f + 100], constants))
# print('({0},{1}):{2:.0f}'.format(x,y,x))
# report_fit(minn.params)
if minn.params['w0'].value not in w:
w.append(minn.params['w0'].value)
s.append(minn.params['sigma'].value)
I.append(minn.params['I'].value)
k.append(minn.params['k'].value)
b.append(minn.params['b'].value)
spectra2 = spectra.copy()
for i in range(len(w)):
xr = np.arange(w[i] - s[i] * 3, w[i] + s[i] * 3, 1, dtype=int)
model = plfunc(w[i], I[i], s[i], 0, 0, xr)
if (spectra2[xr] - model).std() < spectra2[xr].std():
# plot(xr,plfunc(w[i],I[i],s[i],k[i],b[i],xr),'-g',linewidth=2)
spectra2[xr] -= model
noisy_coefs = pywt.wavedec(spectra2, 'db8', level=11, mode='per')
sigma = stand_mad(noisy_coefs[-1])
uthresh = 2 * sigma * np.sqrt(2 * np.log(len(spectra2)))
denoised = noisy_coefs[:]
denoised[1:] = (pywt.threshold(i, uthresh) for i in denoised[1:])
signal = pywt.waverec(denoised, 'db8', mode='per')
cube[:, x, y] = spectra - signal
plt.plot(xx, data)
plt.title('Original Signal')
plt.show()
# 调用wavedec()多级分解函数对数据进行小波变换
# mode指定了数据补齐的方式
#‘per’指周期延拓数据
# 分解层数为9
wavelet_coefs = pywt.wavedec(data, wavtag, level=3, mode='per')
# 进行阈值value构造
# 进行软阈值去噪
# cA9为细节系数
# cD9, cD8, cD7, cD6, cD5, cD4, cD3, cD2, cD1 为模糊系数
# denoised_coefs格式为[array([...]),array([...]),...,array([...])]
sigma = stand_mad(wavelet_coefs[-1])
uthresh = sigma * np.sqrt(2 * np.log(len(data)))
denoised_coefs = wavelet_coefs[:]
denoised_coefs[1:] = (pywt.threshold(coefs, value=uthresh, mode='soft') for coefs in denoised_coefs[1:])
#cA9, cD9, cD8, cD7, cD6, cD5, cD4, cD3, cD2, cD1 = denoised_coefs
# 将去噪得到的系数进行处理，以方便存入txt
# 首先将denoised_coefs中每个array转换tolist为列表，在append入noisy_coefs_list
# 再将noisy_coefs_list中每个列表的float数据转换为str并以“ ”分隔
# 将i_str进行LZW压缩
# 返回的aIn为list，将其中每个0~256整数做参数，返回一个对应字符并拼接
wavelet_coefs_list=[]
for i in denoised_coefs:# i为array
wavelet_coefs_list.append(i.tolist())
storeData=''
i_strAll=''