def asinhScale(data, nonlin, shift, minCut=None, maxCut=None, fname="", rgb=False, fscale=True):
print "Enter asinhScale.............................."
minX=data.min()
maxX=data.max()
if minCut == None:
minCut=minX
if maxCut == None:
maxCut=maxX
output = numpy.array(data, copy=True)
fact=numpy.arcsinh((maxCut-minCut)/nonlin)
print "factor:",fact
output = output + shift
lowCut = numpy.where(output < minCut)
data_i = numpy.where((output > minCut) & (output < maxCut))
hiCut = numpy.where(output > maxCut)
# Zero out low end to avoid negatives
output[lowCut] = 0.0
# perform asinh and scaling to 0.0-1.0
if fscale is True:
output[data_i] = numpy.arcsinh(((output[data_i])/nonlin))/fact
else:
output[data_i] = numpy.arcsinh(((output[data_i])/nonlin))
# Cut off high end
output[hiCut] = 1.0
if rgb is True:
rgbImg = numpy.zeros((output.shape[0], output.shape[1], 3), dtype=float)
# Make RGB gray scale
rgbImg[data_i[0],data_i[1],0] = output[data_i]
rgbImg[data_i[0],data_i[1],1] = output[data_i]
rgbImg[data_i[0],data_i[1],2] = output[data_i]
# Add in lows in Blue and highs in Red
# lows are set to 0.0 and therefore wont show up.
# to make low end visible uncomment:
#output[lowCut] = 1.0
rgbImg[lowCut[0],lowCut[1],2] = output[lowCut]
rgbImg[hiCut[0],hiCut[1],0] = output[hiCut]
# Write out image
if fname != "":
imagetools.imwrite(rgbImg, fname+"_RGB_"+(str(nonlin)+'_'+str(shift)+'_'+str(minCut)+'_'+str(maxCut))+".png")
print "Leaving asinhScale.........................RGB"
return numpy.array(rgbImg, copy=True)
else:
print "out min/max", output.min(), output.max()
# Write out image
if fname != "":
imagetools.imwrite(output, fname+"_"+(str(nonlin)+'_'+str(shift)+'_'+str(minCut)+'_'+str(maxCut))+".png")
print "Leaving asinhScale.........................GRAY"
return numpy.array(output, copy=True)
# multiplicative update is used which preserves positivity
factor_gpu = F.cnvtp(mask_gpu*y_gpu)/(F.cnvtp(mask_gpu*F.cnv(x_gpu))+tol)
gputools.cliplower_GPU(factor_gpu, tol)
x_gpu = x_gpu * factor_gpu
x_max = x_gpu.get()[sf[0]:-sf[0],sf[1]:-sf[1]].max()
gputools.clipupper_GPU(x_gpu, x_max)
del factor_gpu
del F
# ------------------------------------------------------------------------
# For backup intermediate results
# ------------------------------------------------------------------------
if backup:
# Write intermediate results to disk incl. input
imagetools.imwrite(y_gpu.get(), yname(i))
# Crop image to input size
xi = x_gpu.get()[sf2[0]:-sf2[0],sf2[1]:-sf2[1]] / x_max
imagetools.imwrite(xi, xname(i))
# Concatenate PSF kernels for ease of visualisation
f = imagetools.gridF(fs,csf)
imagetools.imwrite(f/f.max(), fname(i))
# ------------------------------------------------------------------------
# For displaying intermediate results
# ------------------------------------------------------------------------
if np.mod(i,1) == 0 and doshow:
pl.figure(1)
factor_gpu = F.cnvtp(
mask_gpu * y_gpu) / (F.cnvtp(mask_gpu * F.cnv(x_gpu)) + tol)
gputools.cliplower_GPU(factor_gpu, tol)
x_gpu = x_gpu * factor_gpu
x_max = x_gpu.get()[sf[0]:-sf[0], sf[1]:-sf[1]].max()
gputools.clipupper_GPU(x_gpu, x_max)
del factor_gpu
del F
# ------------------------------------------------------------------------
# For backup intermediate results
# ------------------------------------------------------------------------
if backup:
# Write intermediate results to disk incl. input
imagetools.imwrite(y_gpu.get(), yname(i))
# Crop image to input size
xi = x_gpu.get()[sf2[0]:-sf2[0], sf2[1]:-sf2[1]] / x_max
imagetools.imwrite(xi, xname(i))
# Concatenate PSF kernels for ease of visualisation
f = imagetools.gridF(fs, csf)
imagetools.imwrite(f / f.max(), fname(i))
# ------------------------------------------------------------------------
# For displaying intermediate results
# ------------------------------------------------------------------------
if np.mod(i, 1) == 0 and doshow:
pl.figure(1)
pl.subplot(121)
def asinhScale(data,
nonlin,
shift,
minCut=None,
maxCut=None,
fname="",
rgb=False,
fscale=True):
print "Enter asinhScale.............................."
minX = data.min()
maxX = data.max()
if minCut == None:
minCut = minX
if maxCut == None:
maxCut = maxX
output = numpy.array(data, copy=True)
fact = numpy.arcsinh((maxCut - minCut) / nonlin)
print "factor:", fact
output = output + shift
lowCut = numpy.where(output < minCut)
data_i = numpy.where((output > minCut) & (output < maxCut))
hiCut = numpy.where(output > maxCut)
# Zero out low end to avoid negatives
output[lowCut] = 0.0
# perform asinh and scaling to 0.0-1.0
if fscale is True:
output[data_i] = numpy.arcsinh(((output[data_i]) / nonlin)) / fact
else:
output[data_i] = numpy.arcsinh(((output[data_i]) / nonlin))
# Cut off high end
output[hiCut] = 1.0
if rgb is True:
rgbImg = numpy.zeros((output.shape[0], output.shape[1], 3),
dtype=float)
# Make RGB gray scale
rgbImg[data_i[0], data_i[1], 0] = output[data_i]
rgbImg[data_i[0], data_i[1], 1] = output[data_i]
rgbImg[data_i[0], data_i[1], 2] = output[data_i]
# Add in lows in Blue and highs in Red
# lows are set to 0.0 and therefore wont show up.
# to make low end visible uncomment:
#output[lowCut] = 1.0
rgbImg[lowCut[0], lowCut[1], 2] = output[lowCut]
rgbImg[hiCut[0], hiCut[1], 0] = output[hiCut]
# Write out image
if fname != "":
imagetools.imwrite(
rgbImg,
fname + "_RGB_" + (str(nonlin) + '_' + str(shift) + '_' +
str(minCut) + '_' + str(maxCut)) + ".png")
print "Leaving asinhScale.........................RGB"
return numpy.array(rgbImg, copy=True)
else:
print "out min/max", output.min(), output.max()
# Write out image
if fname != "":
imagetools.imwrite(
output,
fname + "_" + (str(nonlin) + '_' + str(shift) + '_' +
str(minCut) + '_' + str(maxCut)) + ".png")
print "Leaving asinhScale.........................GRAY"
return numpy.array(output, copy=True)
