def countStepByZcross(fname, look, zcTh, varTh=1, debug=False):
lpf=LPF()
if os.path.isdir(fname):
paths=glob(os.path.join(fname, '*.xml') )
else:
paths=[fname]
steps={}
for path in paths:
d=parseXml2dic(path)
if look==0:
accWf=getAccWf(d)
data=accWf[2]
elif look==1:
data=(d[Keys.kAx]**2+d[Keys.kAy]**2+d[Keys.kAz]**2)**0.5
#data=lpf.lpfScipy(data)
data=lpf.lpfTest(data)
s=countStep(data, zcTh, varTh, debug)
steps[os.path.basename(path)]=s
return steps
pass
def checkNoise(noiseFactor):
mses = []
noises = []
for coeffs_num in coeffs_nums:
sums = []
for i in xrange(30):
noise = map(add, np.random.normal(0, (amplitude / noiseFactor) ** 2, len(list)), list)
clean = LPF.clean(noise, coeffs_num)
sums.append(me(list, clean))
noises.append(me(list, noise))
mses.append(np.mean(sums))
plt.plot(coeffs_nums, mses, label="amplitude/noise_std=" + str(factor))
m = np.mean(noises)
plt.plot(coeffs_nums, [m] * len(coeffs_nums), label="noise reference for " + str(factor))
def stitchByOrder(parts):
byOrder = copy.copy(parts[0])
averaged = copy.copy(byOrder)
for part in parts[1:]:
byOrder, averaged = appendFrac(byOrder, part, averaged)
cleaned, cleaned_time = LPF.clean(averaged)
plt.figure()
plt.title('Stitching by order')
plt.plot(xrange(len(input)), input, color='blue', label='original, SM='+str(cost(input)))
plt.plot(xrange(len(byOrder)), byOrder, color='black', label='byOrder, SM='+str(cost(byOrder)))
plt.plot(xrange(len(averaged)), averaged, color='red', label='byOrder and averaged, SM='+str(cost(averaged)))
plt.plot(xrange(len(cleaned)), cleaned, color='green', label='byOrder and LPF and averaged, SM='+str(cost(cleaned)))
plt.legend().draggable()
def stitch(parts, m1=MEAN_COEFF, m2=STD_COEFF, epsilon=EPSILON_FACTOR, gap=GAP_FACTOR, overlap=OVERLAP_FACTOR):
"""
MEAN_COEFF=m1
STD_COEFF=m2
EPSILON_FACTOR=epsilon
OVERLAP_FACTOR=overlap
GAP_FACTOR=gap
"""
#greedy
parts.sort()
parts.reverse()
averaged = copy.copy(parts[0])
notUsed = []
for part in parts[1:]:
lenBefore = len(averaged)
#greedy, kuku = appendFrac(greedy, part, averaged)
averaged = appendFracaveraged(averaged, part, m1, m2, epsilon, gap, overlap)
if(lenBefore == len(averaged)):
notUsed.append(part)
counter = 0
for part in notUsed:
lenBefore = len(averaged)
averaged = prependFrac(averaged, part, m1, m2, epsilon, gap, overlap)
if(len(averaged) != lenBefore):
counter+=1
notUsed.remove(part)
"""
for part in notUsed:
a = np.sum(averaged)
lcs.lcs(averaged, part, isEqual)
#if(a!=np.sum(averaged)):
# print 'kuku'
"""
clean, clean_time = LPF.clean(averaged)
plt.figure()
plt.plot(clean)
return clean
if not start:
start=True
axvline(i+1-winsz*2, c='r', lw=2)
steps+=getZccByZline(buf, zline, i+1-winsz*2, debug=True, begin=0, end=winsz)
del cbufv[:winsz-overlap]
del cbufvlpf[:winsz-overlap]
i+=1
return steps
#fname=r'D:/Documents/Desktop/step-counting-data/old-linear/segmented/ZC30fast_a9_0.xml'
figure()
d=parseXml2dic(fname)
data=(d[Keys.kAx]**2+d[Keys.kAy]**2+d[Keys.kAz]**2)**0.5
lpf=LPF()
dlpf=lpf.lpfTest(data)
d11=lpf.lpfTest(data, 11)
tslist=d[Keys.kTs]
#plot(tslist, data)
#plot(data, 'b', dlpf, 'r')
#plot(d11, 'g')
plot(dlpf, 'r')
#diff(tslist)
for j in range(len(tslist)):
s2=dyZcrossOnline(data[j], tslist[j], True)
#plot(tmp, 'c')
print "online:", s2
import numpy as np import numpy.random as rnd import matplotlib.pyplot as plt from utils.vicon.amcParser import getAMCperiod from utils.stitching.stitching import MAXIMA_ORDER, CLUSTER_COEFF, plotParts, createParts import utils.stitching.stitching as loop import utils.periodAnalysisUtils as ut from operator import add, sub import utils.LPF as LPF file = 'AMCs/598.amc' joint = 'root' list = getAMCperiod(joint, file) list = list[:538] #stride = list[112:251] #list = ut.alignByMax(stride) #list = ut.alignByMax(list) amplitude = np.max(list) - np.min(list) parts = createParts(list, True, 9, amplitude/10) merged = loop.stitch(parts) merged.sort() print(len(merged)) #out = ut.alignByMax(merged[-1]) out = LPF.clean(merged[-1]) outOff, listOff = ut.alignByBig(out, list) plt.figure() plt.plot(xrange(listOff, listOff+len(list)), list) plt.plot(xrange(outOff, outOff+len(out)), out) plt.show()
list = ut.alignByMax(stride)
amplitude = np.max(list) - np.min(list)
coeffs_num = 16
window_size = 16
maFactor = 1.2
msesMA = []
msesLPF = []
noises = []
noisesFactors = xrange(2,20)
for noiseFactor in noisesFactors:
sumsLPF = []
sumsMA = []
sumsN = []
for i in xrange(30):
noise = map(add, np.random.normal(0,(amplitude/noiseFactor)**2,len(list)), list)
clean = LPF.clean(noise, coeffs_num)
ma = movingAverage(noise, window_size, maFactor)
sumsLPF.append(me(list, clean))
sumsMA.append(me(list, ma))
sumsN.append(me(list, noise))
msesMA.append(np.mean(sumsMA))
msesLPF.append(np.mean(sumsLPF))
noises.append(np.mean(sumsN))
plt.plot(noisesFactors, msesMA, label='MA' )
plt.plot(noisesFactors, msesLPF, label='LPF' )
plt.plot(noisesFactors, noises, label='noise reference for ')
plt.xlabel('noises factors')
plt.ylabel('me')
plt.title('Mean error of different noises')
plt.legend().draggable()
