def vector_subphog(X, Y, W, H, grayscale, grayD):
#from numpy.ma import MaskedArray
#import scipy.stats
subhistogram = numpy.zeros(PHOG_BINS, dtype="double")
subphog = grayscale[X:(X+W), Y:(Y+H)] < 50
#bindex = numpy.floor((MaskedArray(grayD, mask=subphog) / numpy.pi + 0.5) * PHOG_BINS).astype('int')
# bins = numpy.bincount(numpy.where(bindex == PHOG_BINS, 0, bindex))
print_array_info(grayD, title="grayD")
print_array_info(subphog, title="subphog")
#print_array_info(bindex, title="bindex")
subdirectionals = (grayD[subphog] / numpy.pi + 0.5) * PHOG_BINS
histo = numpy.bincount(
numpy.where(
subdirectionals.astype('int') == PHOG_BINS,
0, subdirectionals.astype('int')))
print_array_info(histo, title="HISTOGRAM (grayD[subphog])")
# print_array_info(bins, title="BINS")
# print("BINDEXES:")
# print(numpy.array_repr(bindexes, max_line_width=200))
# print("max: %s min: %s" % (numpy.max(bindexes), numpy.min(bindexes)))
#return
# yo = numpy.where(numpy.floor(bindexfp) == PHOG_BINS, 0, numpy.floor(bindexfp))
# bindexes = numpy.where(bindexfp == numpy.floor(bindexfp),
# (numpy.floor(bindexfp).astype('int'), 1),
# (numpy.floor(bindexfp).astype('int'), bindexfp - numpy.floor(bindexfp)))
# bindexes = numpy.zeros_like(subphog)
# bindexes += numpy.where(values == numpy.floor(values).astype('int'),
# 1, (values - numpy.floor(values)))
# bindexes += numpy.where(values == numpy.ceil(values).astype('int'),
# 0, (numpy.ceil(values) - values))
# print("BINDEXES:")
# print(bindexes)
# return
# "normalize histogram to max norm."
# NB. this piece is actually vectorized
histomax = numpy.max(subhistogram)
if histomax > 0.0:
subhistogram = numpy.minimum(
QUANTIZATION_FACTOR,
numpy.floor(
QUANTIZATION_FACTOR * subhistogram / histomax))
return subhistogram
def timetest_cythonized_HOG(ndim):
hogg = hog(ndim)
print("Cythonized HOG() function:")
print_array_info(hogg)
print("")
