#%%

import svgwrite

import os

import cairosvg

import tempfile

import skimage.io

import skimage.transform

import subprocess

import numpy

import time

import pickle

import skimage.feature

import sklearn.cross_validation

import sklearn.svm

os.chdir('/home/bbales2/microstructure_python')

import microstructure.features

os.chdir('/home/bbales2/lukestuff')

#%%

def get_features(im):

return hist

filenames = [('scaled_rafting/ReneN5_0pct_10000x_Transverse_DendriteCore.TIF.png', 0.0),

('scaled_rafting/ReneN5_2pct_10000x_Transverse_DendriteCore.TIF.png', 2.0),

('scaled_rafting/ReneN5_5pct_10000x_Transverse_DendriteCore.TIF.png', 5.0)]

trainingx = []

trainingy = []

testingx = []

testingy = []

ims = []

trains = []

tests = []

for filename, y in filenames:

im = skimage.io.imread(filename, as_grey = True).astype('double')

im -= numpy.mean(im.flatten())

im /= numpy.std(im.flatten())

print im.shape

ims.append(im)

train = im[:3 * im.shape[0] / 4, :]

test = skimage.filters.gaussian_filter(im[3 * im.shape[0] / 4:, :], 1.0)#sklearn.cross_validation.train_test_split(list(im))

train = numpy.array(train)

test = numpy.array(test)

trainx = get_features(train)

testx = get_features(test)

trains.append(train)

tests.append(test)

trainy = numpy.ones((trainx.shape[0], trainx.shape[1])) * y

testy = numpy.ones((testx.shape[0], testx.shape[1])) * y

trainingx.extend(trainx.reshape((-1, trainx.shape[-1])))

trainingy.extend(trainy.flatten())

testingx.extend(testx.reshape((-1, testx.shape[-1])))

testingy.extend(testy.flatten())

trainingx = numpy.array(trainingx)

trainingy = numpy.array(trainingy)

testingx = numpy.array(testingx)

testingy = numpy.array(testingy)

print trainingx.shape

print trainingy.shape

print testingx.shape

print testingy.shape

idxs = range(len(trainingx))

numpy.random.shuffle(idxs)

trainingx = trainingx[idxs]

trainingy = trainingy[idxs]

plt.imshow(test)

plt.show()

plt.imshow(train)

plt.show()

#%%

svm = sklearn.svm.SVR()

Cs = numpy.logspace(0, 2, 8)

gammas = numpy.logspace(-6, -4, 10)

accuracies = numpy.zeros((8, 10))

for i, c in enumerate(Cs):

for j, gamma in enumerate(gammas):

svm = sklearn.svm.SVR(C = c, gamma = gamma)

svm.fit(trainingx[:1000], trainingy[:1000])

accuracies[i, j] = svm.score(testingx, testingy)

print i, j

plt.imshow(accuracies, interpolation = 'NONE')

plt.xticks(range(10), ['{:2.2e}'.format(v) for v in gammas])

plt.yticks(range(8), ['{:2.1e}'.format(v) for v in Cs])

plt.gcf().set_size_inches((10, 10))

plt.colorbar()

plt.show()

#%%

svm = sklearn.svm.SVR(C = 5.0, gamma = 1e-5)

svm.fit(trainingx[:1000], trainingy[:1000])

print svm.score(testingx, testingy)

rafting = svm.predict(testingx.reshape((-1, testingx.shape[-1]))).reshape((3 * testx.shape[0], testx.shape[1]))

plt.imshow(rafting)

plt.show()

#%%

svm = sklearn.linear_model.LinearRegression()

svm.fit(trainingx[:1000], trainingy[:1000])

print svm.score(testingx, testingy)

rafting = svm.predict(testingx.reshape((-1, testingx.shape[-1]))).reshape((3 * testx.shape[0], testx.shape[1]))

plt.imshow(rafting)

plt.show()

scores = []

for i in range(0, testingx.shape[1], 1):

fs2 = list(reversed(numpy.argsort(numpy.abs(svm.coef_))))

fs = fs2[:i + 1]

print i, len(fs)

lr2 = sklearn.linear_model.LinearRegression()

lr2.fit(trainingx[:1000, fs], trainingy[:1000])

rafting = lr2.predict(testingx[:, fs]).reshape((3 * testx.shape[0], testx.shape[1]))

plt.imshow(rafting)

plt.title(str(i))

plt.show()

scores.append(lr2.score(testingx[:, fs], testingy))

#print

plt.plot(scores)

plt.show()

#%%

for i in fs2:

print svm.coef_[i], "{0:04b}".format(i)

#%%

for im, (filename, y) in zip(ims, filenames):

#if True:

# im = skimage.io.imread('gtdmix.png', as_grey = True).astype('double')

# im -= numpy.mean(im.flatten())

# im /= numpy.std(im.flatten())

filename = 'gtdmix'

print im.shape

#im2, test = sklearn.cross_validation.train_test_split(list(im), test_size = 0.0)

#im2 = numpy.array(im2)

#print im2.shape

features = get_features(im)

rafting = svm.predict(features.reshape((-1, features.shape[-1]))).reshape((features.shape[0], features.shape[1]))

#plt.imshow(im, interpolation = 'NONE', cmap = plt.cm.gray)

#plt.show()

plt.imshow(im, interpolation = 'NONE', cmap = plt.cm.gray)

plt.imshow(rafting, interpolation = 'NONE', extent = (0, im.shape[1], im.shape[0], 0), alpha = 0.5, vmin = 0.0, vmax = 5.0)

plt.title(filename)

plt.gcf().set_size_inches((17, 10))

plt.colorbar()

plt.show()

#plt.imshow(rafting, interpolation = 'NONE', vmin = 0.0, vmax = 5.0)

#plt.show()

#%%

f0 = get_features(im)

f1 = get_features(im2)