def get_change_point(self, other):
seq = range(256)
seq.reverse()
max = -1
max_x = 0
for kc in seq:
xc = BinaryImage.to_xc(kc)
pb = self.binarization(xc)
qb = other.binarization(xc)
d = BinaryImage(pb).diff(qb)
dd = DensityDistribution(d)
black = DensityDistribution(Image.new('L', self.image.size, 0))
divergence = dd.divergence(black)
if abs(divergence) > max:
max = abs(divergence)
max_x = xc
return max_x
# -*- coding: utf-8 -*-
#
# assignment5.output
# created by 25090335 Kohki Miki on 2011/08/11
#
from PIL import Image
from assignment4.no1 import DensityDistribution
from matplotlib import pyplot
image = DensityDistribution(Image.open(r"../../Resources/CIMG0211.gif"))
xs, ys = image.density_distribution()
for x, y in zip(xs, ys):
print x, y
class GaussianFitting(object):
def gaussian(self, params, x):
a, m1, m2, s1, s2 = params
p = a/numpy.sqrt(2*numpy.pi*s1**2)*numpy.exp(-(x-m1)**2/(2*s1**2)) + (1-a)/numpy.sqrt(2*numpy.pi*s2**2)*numpy.exp(-(x-m2)**2/(2*s2**2))
return p
def residuals(self, params, x, y):
err = self.gaussian(params, x) - y
return err
def fitting(self, x, y, initial=None):
if not initial: initial = [0.] * 5
result = leastsq(self.residuals, initial, args=(x, y), full_output=False)
return result
if __name__ == '__main__':
p1 = DensityDistribution(Image.open(r"../../Resources/CIMG0209.gif"))
x, y = p1.density_distribution()
gf = GaussianFitting()
initial = (0.63583867, 3.22051539, 1.26030523, 0.64014307, 0.52168001)
params = gf.fitting(x, y, initial)
time = numpy.linspace(x[0], x[-1], 100)
print params[0]
pyplot.plot(x, y, x, [gf.gaussian(params[0], t) for t in x], "r-")
pyplot.title(r"CIMG0209.gif")
pyplot.xlabel("x")
pyplot.ylabel("P(x)")
pyplot.show()
p2 = DensityDistribution(Image.open(r"../../Resources/CIMG0210.gif"))
x, y = p2.density_distribution()
gf = GaussianFitting()