def genLMS(spectrum, filters, fundamental='neitz', LMSpeaks=[559.0, 530.0, 419.0],
remove_filters=True, normalize=True):
'''
'''
if len(LMSpeaks) != 3:
print 'LMSpeaks must be length 3! Using defaults: 559, 530, 417nm'
LMSpeaks = [559.0, 530.0, 421.0]
minLam = np.min(spectrum)
maxLam = np.max(spectrum)
step = spectrum[1] - spectrum[0]
if fundamental.lower() == 'stockman':
sens = ss.stockmanfund(minLambda=minLam, maxLambda=maxLam, resolution=step)
LS = sens[:, 0]
MS = sens[:, 1]
SS = sens[:, 2]
Lresponse = LS * spectrum
Mresponse = MS * spectrum
Sresponse = SS * spectrum
elif fundamental.lower() == 'stockspecsens':
sens = ss.stockman(minLambda=minLam, maxLambda=maxLam, resolution=step)
LS = sens[:, 0]
MS = sens[:, 1]
SS = sens[:, 2]
Lresponse = LS / filters * spectrum
Mresponse = MS / filters * spectrum
Sresponse = SS / filters * spectrum
elif fundamental.lower() == 'neitz':
LS = ss.neitz(LMSpeaks[0], 0.5, False, minLam,
maxLam, step)
MS = ss.neitz(LMSpeaks[1], 0.5, False, minLam,
maxLam, step)
SS = ss.neitz(LMSpeaks[2], 0.4, False, minLam,
maxLam, step)
Lresponse = LS / filters * spectrum
Mresponse = MS / filters * spectrum
Sresponse = SS / filters * spectrum
# normalize to peak at unity
Lnorm = Lresponse / np.max(Lresponse)
Mnorm = Mresponse / np.max(Mresponse)
Snorm = Sresponse / np.max(Sresponse)
if remove_filters and normalize:
return Lnorm, Mnorm, Snorm
if remove_filters and not normalize:
return Lresponse, Mresponse, Sresponse
if not remove_filters:
return LS, MS, SS
def genTetraConvMatrix(self, Xpeak):
'''
'''
self.setLights(self.param['stim'])
minspec = min(self.spectrum)
maxspec = max(self.spectrum)
Xsens = ss.neitz(Xpeak, 0.5, False, minspec,
maxspec, 1)
Xresponse = Xsens / self.filters * self.spectrum
Xnorm = Xresponse / np.max(Xresponse)
lights = {'l': 600, 'm': 510, 's': 420, 'x': 720}
convMatrix = np.array([
[np.interp(lights['l'], self.spectrum, self.Lnorm),
np.interp(lights['m'], self.spectrum, self.Lnorm),
np.interp(lights['s'], self.spectrum, self.Lnorm),
np.interp(lights['x'], self.spectrum, self.Lnorm)],
[np.interp(lights['l'], self.spectrum, self.Mnorm),
np.interp(lights['m'], self.spectrum, self.Mnorm),
np.interp(lights['s'], self.spectrum, self.Mnorm),
np.interp(lights['x'], self.spectrum, self.Mnorm)],
[np.interp(lights['l'], self.spectrum, self.Snorm),
np.interp(lights['m'], self.spectrum, self.Snorm),
np.interp(lights['s'], self.spectrum, self.Snorm),
np.interp(lights['x'], self.spectrum, self.Snorm)],
[np.interp(lights['l'], self.spectrum, Xnorm),
np.interp(lights['m'], self.spectrum, Xnorm),
np.interp(lights['s'], self.spectrum, Xnorm),
np.interp(lights['x'], self.spectrum, Xnorm)]])
return convMatrix
def get_cones(self):
'''
'''
L_cones = spect.neitz(LambdaMax=self.cones['L_peak'],
OpticalDensity=self.cones['L_OD'],
LOG=False, StartWavelength=self.spectrum[0],
EndWavelength=self.spectrum[-1],
resolution=1, EXTINCTION=False)
M_cones = spect.neitz(LambdaMax=self.cones['M_peak'],
OpticalDensity=self.cones['M_OD'],
LOG=False, StartWavelength=self.spectrum[0],
EndWavelength=self.spectrum[-1],
resolution=1, EXTINCTION=False)
M = M_cones * self.lens
self.cones['M'] = M / M.max()
L = L_cones * self.lens
self.cones['L'] = L / L.max()
def genReceptiveFields(freqs, cone_spacing, peaks=None, wvlen=None):
"""Create a difference of gaussians receptive field and plot it.
:param excite_SD: standard deviation parameter of excitatory gaussian.
:param inhibit_SD: standard deviation parameter of inhibitory \
surround gaussian. Default = 5.0.
:returns: RF_DOG, RF_SPLINE, FFT_RF
"""
# traditional RF
N = 300
Xvals = np.arange(-15, 15, 10. / (2.0 * N))
RF_DOG = []
FFT_RF = []
RField = []
if peaks is not None and len(peaks) == 2 and wvlen is not None:
K_excite = ss.neitz(LambdaMax=peaks[0], OpticalDensity=0.5, LOG=False,
StartWavelength=wvlen, EndWavelength=wvlen,
resolution=1, EXTINCTION=False)
K_inhibit = ss.neitz(LambdaMax=peaks[1], OpticalDensity=0.5, LOG=False,
StartWavelength=wvlen, EndWavelength=wvlen,
resolution=1, EXTINCTION=False)
else:
K_excite, K_inhibit = 1, 1
RF_DOG = findSpacing(Xvals, K_excite, K_inhibit, cone_spacing)
FFT_RF = Fourier(RF_DOG, N)
length = np.floor(FFT_RF.shape[0] / 2.) + 1
RField = interpRField(freqs, Xvals[length:] * 60,
FFT_RF[length:])
receptive_field = {
'length': length,
'dog': RF_DOG,
'coneResponse': FFT_RF,
'fft': RField,
'xvals': Xvals,
}
return receptive_field
def genFirstStage(self, startLambda=390, endLambda=750, step=1,
Out='anti-log', OD=None):
'''Compute the first stage in the model.
'''
if OD is None:
OD = [0.4, 0.38, 0.33]
lambdas = np.arange(startLambda, endLambda + step, step)
L_cones = ss.neitz(LambdaMax=self.maxSens['l'], LOG=False,
StartWavelength=startLambda,
OpticalDensity=OD[0],
EndWavelength=endLambda,
resolution=step)
L_cones /= self.lensMacula * lambdas / lambdas[-1]
M_cones = ss.neitz(LambdaMax=self.maxSens['m'], LOG=False,
StartWavelength=startLambda,
OpticalDensity=OD[1],
EndWavelength=endLambda,
resolution=step)
M_cones /= self.lensMacula * lambdas / lambdas[-1]
S_cones = ss.neitz(LambdaMax=self.maxSens['s'], LOG=False,
StartWavelength=startLambda,
OpticalDensity=OD[2],
EndWavelength=endLambda,
resolution=step)
S_cones /= self.lensMacula * lambdas / lambdas[-1]
self.FirstStage = {
'lambdas': lambdas,
'wavelen': {'startWave': startLambda, 'endWave': endLambda,
'step': step, },
'L_cones': L_cones,
'M_cones': M_cones,
'S_cones': S_cones,
}
def diff_sens(self, curve):
'''
'''
# maximum possible
fake_cone = spect.neitz(
LambdaMax=self.spectrum[self.LED['ind0']],
OpticalDensity=0.3,
LOG=False,
StartWavelength=self.spectrum[0],
EndWavelength=self.spectrum[-1],
resolution=1, EXTINCTION=False)
norm = (np.sum(fake_cone * self.LED['ref']) /
np.sum(curve * self.LED['ref']))
return norm
def test_neitz(self):
sens = sp.neitz(LambdaMax=559, OpticalDensity=0.5, LOG=False,
StartWavelength=390, EndWavelength=770, resolution=1)
self.assertTrue(len(sens) == 381)
self.assertTrue(round(np.max(sens), 2) == 1.0)
