def prewitt_ndimage(arr):
UINT8 = np.iinfo(np.uint8)
x = ndimage.prewitt(arr, axis=0)
y = ndimage.prewitt(arr)
temp = np.hypot(x, y).clip(UINT8.min, UINT8.max).astype(np.uint8)
set_border(temp, 255)
return temp
def test_multiple_modes():
# Test that the filters with multiple mode cababilities for different
# dimensions give the same result as applying a single mode.
arr = np.array([[1., 0., 0.],
[1., 1., 0.],
[0., 0., 0.]])
mode1 = 'reflect'
mode2 = ['reflect', 'reflect']
assert_equal(sndi.gaussian_filter(arr, 1, mode=mode1),
sndi.gaussian_filter(arr, 1, mode=mode2))
assert_equal(sndi.prewitt(arr, mode=mode1),
sndi.prewitt(arr, mode=mode2))
assert_equal(sndi.sobel(arr, mode=mode1),
sndi.sobel(arr, mode=mode2))
assert_equal(sndi.laplace(arr, mode=mode1),
sndi.laplace(arr, mode=mode2))
assert_equal(sndi.gaussian_laplace(arr, 1, mode=mode1),
sndi.gaussian_laplace(arr, 1, mode=mode2))
assert_equal(sndi.maximum_filter(arr, size=5, mode=mode1),
sndi.maximum_filter(arr, size=5, mode=mode2))
assert_equal(sndi.minimum_filter(arr, size=5, mode=mode1),
sndi.minimum_filter(arr, size=5, mode=mode2))
assert_equal(sndi.gaussian_gradient_magnitude(arr, 1, mode=mode1),
sndi.gaussian_gradient_magnitude(arr, 1, mode=mode2))
assert_equal(sndi.uniform_filter(arr, 5, mode=mode1),
sndi.uniform_filter(arr, 5, mode=mode2))
def compute_gradients(img):
"""
Computes the gradients of the input image in the x and y direction using a
differentiation filter.
##########################################################################
# TODO: Design a differentiation filter and update the docstring. Stick #
# to a pure differentiation filter for this assignment. #
# Hint: Look at Slide 14 from Lecture 3: Gradients. #
##########################################################################
Input: Grayscale image of shape (H x W)
Outputs: gx, gy gradients in x and y directions respectively
"""
gray = np.copy(img)
if (len(gray) == 3):
gray = rgb2gray(gray)
# gx = gy = np.zeros_like(img)
gx = -ndimage.prewitt(gray, axis=1, mode='constant')
gy = -ndimage.prewitt(gray, axis=0, mode='constant')
##########################################################################
# TODO: Design a pure differentiation filter and use correlation to #
# compute the gradients gx and gy. You might have to try multiple #
# filters till the test below passes. All the tests after will fail if #
# this one does not pass. #
##########################################################################
return gx, gy
def findedges(image, image_file, issave):
data = image[y0:y1, x0:x1, :]
x, y = np.meshgrid(np.arange(data.shape[1]), np.arange(data.shape[0]))
gaus = scimg.fourier_gaussian(data[:, :, 0], sigma=0.01)
can_x = scimg.prewitt(gaus, axis=0)
can_y = scimg.prewitt(gaus, axis=1)
can = np.hypot(can_x, can_y)
# pulling out object edges
fig3, ax3 = plt.subplots(2, 1, figsize=(10, 7))
ax3[0].pcolormesh(x, y, can, cmap='gist_ncar')
bin_size = 30 # total bins to show
percent_cutoff = 0.05 # cutoff once main peak tapers to 5% of max
hist_vec = np.histogram(can.ravel(), bins=bin_size)
hist_x, hist_y = hist_vec[0], hist_vec[1]
for ii in range(np.argmax(hist_x), bin_size):
hist_max = hist_y[ii]
if hist_x[ii] < percent_cutoff * np.max(hist_x):
break
# scatter points where objects exist
ax3[1].plot(x[can > hist_max],
y[can > hist_max],
marker='.',
linestyle='',
label='Scatter Above 5% Dropoff')
ax3[1].set_xlim(np.min(x), np.max(x))
ax3[1].set_ylim(np.min(y), np.max(y))
ax3[1].legend()
if issave:
plt.savefig(image_file.split('.')[0] + '_edges.png')
else:
plt.show()
print(hist_vec[1])
return stat_dsp(hist_vec[1], threshold)
def run(self, ips, snap, img, para=None):
nimg.prewitt(snap,
axis={
'horizontal': 0,
'vertical': 1
}[para['axis']],
output=img)
def test_multiple_modes():
# Test that the filters with multiple mode cababilities for different
# dimensions give the same result as applying a single mode.
arr = np.array([[1., 0., 0.],
[1., 1., 0.],
[0., 0., 0.]])
mode1 = 'reflect'
mode2 = ['reflect', 'reflect']
assert_equal(sndi.gaussian_filter(arr, 1, mode=mode1),
sndi.gaussian_filter(arr, 1, mode=mode2))
assert_equal(sndi.prewitt(arr, mode=mode1),
sndi.prewitt(arr, mode=mode2))
assert_equal(sndi.sobel(arr, mode=mode1),
sndi.sobel(arr, mode=mode2))
assert_equal(sndi.laplace(arr, mode=mode1),
sndi.laplace(arr, mode=mode2))
assert_equal(sndi.gaussian_laplace(arr, 1, mode=mode1),
sndi.gaussian_laplace(arr, 1, mode=mode2))
assert_equal(sndi.maximum_filter(arr, size=5, mode=mode1),
sndi.maximum_filter(arr, size=5, mode=mode2))
assert_equal(sndi.minimum_filter(arr, size=5, mode=mode1),
sndi.minimum_filter(arr, size=5, mode=mode2))
assert_equal(sndi.gaussian_gradient_magnitude(arr, 1, mode=mode1),
sndi.gaussian_gradient_magnitude(arr, 1, mode=mode2))
assert_equal(sndi.uniform_filter(arr, 5, mode=mode1),
sndi.uniform_filter(arr, 5, mode=mode2))
def run(self, ips, snap, img, para = None):
if para['axis']=='both':
img[:] = np.abs(nimg.prewitt(snap, axis=0, output=img.dtype))
img += np.abs( nimg.prewitt(snap, axis=1, output=img.dtype))
else:
nimg.prewitt(snap, axis={'horizontal':0,'vertical':1}[para['axis']], output=img)
img[:] = np.abs(img)
img //= 3
def prewitt(img):
"""Function to apply a prewitt filter on a given input image.
:param img: {numpy.array} image as numpy array
:return: {numpy.array} filtered image
"""
sx = ndimage.prewitt(img, axis=0, mode='constant')
sy = ndimage.prewitt(img, axis=1, mode='constant')
prew = np.hypot(sx, sy)
return prew
def prewitt(im):
"""
Edge detector using a Prewitt filter
LONG RUN!
Arguments:
im - input image
For classifier - recommended multiple runs after applying gaussian filter w sigma 1.0-16.0
"""
im = im.astype('int32')
sx = ndimage.prewitt(im, axis=(im.ndim - 2), mode='constant')
sy = ndimage.prewitt(im, axis=(im.ndim - 1), mode='constant')
return np.hypot(sx, sy)
def doCompute(): inlpos = xa.SI['nrinl']//2 crlpos = xa.SI['nrcrl']//2 while True: xa.doInput() indata = xa.Input['Input'] xa.Output['In-line gradient'] = prewitt(indata, axis=0)[inlpos,crlpos,:] xa.Output['Cross-line gradient'] = prewitt(indata, axis=1)[inlpos,crlpos,:] xa.Output['Z gradient'] = prewitt(indata, axis=2)[inlpos,crlpos,:] xa.Output['Average Gradient'] = ( xa.Output['In-line gradient'] + xa.Output['Cross-line gradient'] + xa.Output['Z gradient'] )/3 xa.doOutput()
def doCompute():
#
# index of current trace position in Input numpy array
#
ilndx = xa.SI["nrinl"] // 2
crldx = xa.SI["nrcrl"] // 2
while True:
xa.doInput()
xa.Output["In-line gradient"] = prewitt(xa.Input, axis=0)[ilndx, crldx, :]
xa.Output["Cross-line gradient"] = prewitt(xa.Input, axis=1)[ilndx, crldx, :]
xa.Output["Z gradient"] = prewitt(xa.Input, axis=2)[ilndx, crldx, :]
xa.Output["Average gradient"] = (
xa.Output["In-line gradient"] + xa.Output["Cross-line gradient"] + xa.Output["Z gradient"]
) / 3
xa.doOutput()
def genSimpleFeatures(volume):
return [
ndimage.prewitt(volume),
ndimage.sobel(volume)
] + \
genBlurSharpen(volume, 2.0) + \
genBlurSharpen(volume, 5.0)
def edge_filters(self):
''' Plot five edge-filters (kernels) in grayscale
'''
self.gray = rgb2gray(self.im)
self.edges = {
'Original': self.im,
'Grayscale': self.gray,
'Sobel': ndimage.sobel(self.gray),
'Prewitt': ndimage.prewitt(self.gray),
'Laplacian': ndimage.laplace(self.gray, mode='reflect'),
'LoG': ndimage.gaussian_laplace(self.gray, sigma=1, mode='reflect')
}
fig, axes = plt.subplots(2, 3, figsize=(18, 10))
axs = iter(axes.ravel())
for name, edge in self.edges.items():
ax = next(axs)
ax.imshow(edge, cmap='gray')
ax.set_title(name)
fig.tight_layout()
plt.savefig('.'.join(FNAME.split('.')[:-1]) + '_processed.png')
def doCompute():
inlpos = xa.SI['nrinl'] // 2
crlpos = xa.SI['nrcrl'] // 2
while True:
xa.doInput()
indata = xa.Input['Input']
xa.Output['In-line gradient'] = prewitt(indata, axis=0)[inlpos,
crlpos, :]
xa.Output['Cross-line gradient'] = prewitt(indata, axis=1)[inlpos,
crlpos, :]
xa.Output['Z gradient'] = prewitt(indata, axis=2)[inlpos, crlpos, :]
xa.Output['Average Gradient'] = (xa.Output['In-line gradient'] +
xa.Output['Cross-line gradient'] +
xa.Output['Z gradient']) / 3
xa.doOutput()
def doCompute():
#
# index of current trace position in Input numpy array
#
ilndx = xa.SI['nrinl'] // 2
crldx = xa.SI['nrcrl'] // 2
while True:
xa.doInput()
xa.Output['In-line gradient'] = prewitt(xa.Input, axis=0)[ilndx,
crldx, :]
xa.Output['Cross-line gradient'] = prewitt(xa.Input, axis=1)[ilndx,
crldx, :]
xa.Output['Z gradient'] = prewitt(xa.Input, axis=2)[ilndx, crldx, :]
xa.Output['Average gradient'] = (xa.Output['In-line gradient'] +
xa.Output['Cross-line gradient'] +
xa.Output['Z gradient']) / 3
xa.doOutput()
def prewitt_operator(input, threshold=3, axis=2):
input_y = np.ndarray(shape=(np.shape(input)[1:]), dtype=np.float32)
input_x = np.ndarray(shape=(np.shape(input)[1:]), dtype=np.float32)
output = np.ndarray(shape=(np.shape(input)), dtype=np.float32)
for _ in range(len(input)):
input_y[:, :] = ndimage.prewitt(input[_, :, :], 0)
input_x[:, :] = ndimage.prewitt(input[_, :, :], 1)
if axis == 0:
output[_, :, :] = input_y[:, :]
elif axis == 1:
output[_, :, :] = input_x[:, :]
elif axis == 2:
output[_, :, :] = np.sqrt(
np.square(input_x[:, :]) + np.square(input_y[:, :]))
output[_, :, :] = Relu_threshold(output[_, :, :],
threshold=threshold)
return output
def prewit(self):
from scipy.ndimage import prewitt
img = cv2.imread(self.fileName2, 0)
prewit = prewitt(img)
cv2.imwrite('./sonuclar/prewitt.png', prewit)
w, h = self.gvFiltreIslem.width() - 5, self.gvFiltreIslem.height() - 5
self.gvFiltreIslem.setScene(
self.show_image('./sonuclar/prewitt.png', w, h))
def hpfPrewitt(fileName,num):
workData = getData(fileName,num)
preFilter = ndimage.prewitt(workData)
mfSave = Image.fromarray(preFilter)
mfSave = mfSave.convert('1')
mfSave.save('Prewitt Filter.png')
imageGUI.imdisplay('Prewitt Filter.png','Prewitt',1)
def test_multiple_modes_prewitt():
# Test prewitt filter for multiple extrapolation modes
arr = np.array([[1.0, 0.0, 0.0], [1.0, 1.0, 0.0], [0.0, 0.0, 0.0]])
expected = np.array([[1.0, -3.0, 2.0], [1.0, -2.0, 1.0], [1.0, -1.0, 0.0]])
modes = ["reflect", "wrap"]
assert_equal(expected, sndi.prewitt(arr, mode=modes))
def test_multiple_modes_prewitt():
# Test prewitt filter for multiple extrapolation modes
arr = np.array([[1., 0., 0.], [1., 1., 0.], [0., 0., 0.]])
expected = np.array([[1., -3., 2.], [1., -2., 1.], [1., -1., 0.]])
modes = ['reflect', 'wrap']
assert_equal(expected, sndi.prewitt(arr, mode=modes))
def file_controller(in_path, out_path):
""" ---------- pre-process image ---------- """
image = cv2.imread(in_path)
gray = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
blurred = cv2.GaussianBlur(gray, (3, 3), 0)
""" ---------- create cv2 canny detections ---------- """
start = time.time()
cannywide = cv2.Canny(blurred, 10, 200)
end = time.time()
print("canny_wide: " + str(end - start))
start = time.time()
cannytight = cv2.Canny(blurred, 225, 250)
end = time.time()
print("canny_tight: " + str(end - start))
start = time.time()
cannyauto = auto_canny(blurred) # uses auto_canny function with sigma
end = time.time()
print("canny_auto: " + str(end - start))
""" ---------- create laplacian detection ---------- """
start = time.time()
lap = cv2.Laplacian(blurred,cv2.CV_64F)
end = time.time()
print("laplacian: " + str(end - start))
""" ---------- create sobel operations ---------- """
start = time.time()
sobelx = cv2.Sobel(blurred,cv2.CV_64F,1,0,ksize=5) # x
end = time.time()
print("sobel_x: " + str(end - start))
start = time.time()
sobely = cv2.Sobel(blurred,cv2.CV_64F,0,1,ksize=5) # y
end = time.time()
print("sobel_y: " + str(end - start))
""" ---------- create prewitt operator ---------- """
start = time.time()
prewitt = ndi.prewitt(blurred)
end = time.time()
print("prewitt: " + str(end - start))
""" ---------- write detections to files ---------- """
start = time.time()
cv2.imwrite(out_path + '_canny_auto.jpg', cannyauto)
cv2.imwrite(out_path + '_canny_tight.jpg', cannytight)
cv2.imwrite(out_path + '_canny_wide.jpg', cannywide)
cv2.imwrite(out_path + '_laplacian.jpg', lap)
cv2.imwrite(out_path + '_sobel_x.jpg', sobelx)
cv2.imwrite(out_path + '_sobel_y.jpg', sobely)
cv2.imwrite(out_path + '_prewitt.jpg', prewitt)
end = time.time()
print("Writing images out: " + str(end - start))
def on_prewitt_btn_filter_clicked(self):
from scipy.ndimage import prewitt
img = cv2.imread(self.fileName)
prewit = prewitt(img)
cv2.imwrite('./islenen/prewitt.png', prewit)
w, h = self.operationsGV_filterTab.width(
) - 5, self.operationsGV_filterTab.height() - 5
self.operationsGV_filterTab.setScene(
self.show_image('./islenen/prewitt.png', w, h))
def prewitt(in_path, out_path):
""" Take a directory path, writes result to another path """
print("Creating prewitt...")
start = time.time() # timer start
img = image_preprocess(in_path)
p = ndi.prewitt(img)
cv2.imwrite(out_path + '_prewitt.jpg', p)
end = time.time() # timer end
print("total time: " + str(round(end - start, 2)))
def convolute(self, image, filtr):
from scipy import ndimage
if filtr == 'SOBEL':
return ndimage.sobel(image)
if filtr == 'GAUSSIAN':
return ndimage.gaussian_filter(image,sigma=20)
if filtr == 'LAPLACE':
return ndimage.laplace(image)
if filtr == 'UNIFORM':
return ndimage.uniform_filter(image)
if filtr == 'PREWITT':
return ndimage.prewitt(image)
######################################################################
def prewittFilterRGB(inputData):
r = np.zeros(inputData.shape)
g = np.zeros(inputData.shape)
b = np.zeros(inputData.shape)
ndimage.prewitt(inputData, axis=1, output=r)
ndimage.prewitt(inputData, axis=0, output=g)
ndimage.prewitt(inputData, axis=2, output=b)
return GradientCalculation.normalize(np.concatenate((r[...,np.newaxis],g[...,np.newaxis],b[...,np.newaxis]),axis=3))
def test_multiple_modes_prewitt():
# Test prewitt filter for multiple extrapolation modes
arr = np.array([[1., 0., 0.],
[1., 1., 0.],
[0., 0., 0.]])
expected = np.array([[1., -3., 2.],
[1., -2., 1.],
[1., -1., 0.]])
modes = ['reflect', 'wrap']
assert_equal(expected,
sndi.prewitt(arr, mode=modes))
def filter(img):
#Take greyscale image of 256 * 192
result = img
result = nd.sobel(result)
result = nd.prewitt(result)
result = nd.median_filter(result, size=20)
#Eliminate grey areas more than 190 these are white areas
mask1 = result > 190
mask1 = (mask1 != True)
mask1 = mask1 * 1
result = result * mask1
#Return greyscale image of 256 * 192
img = Image.fromarray(result)
return img.convert('L')
def crFind(img, var, nsig=10., sigfrac=0.3):
simg = img / var**0.5
deriv = ndimage.prewitt(abs(simg))
deriv = ndimage.sobel(deriv)
mean, std = clip(deriv[deriv != 0.])
crmask = numpy.where(abs(deriv) > 15 * std, 1, 0)
crmask = ndimage.maximum_filter(crmask, 3)
#crmask = ndimage.minimum_filter(crmask,3)
crmask = numpy.where((crmask == 1) & (simg > 5), 1, 0)
return crmask
thresh = nsig * var**0.5
n = 5
blkimg = img.repeat(n, 0).repeat(n, 1)
deriv = ndimage.laplace(blkimg) * -1.
deriv[deriv < 0] = 0.
d = iT.resamp(deriv, n)
m = numpy.where((d > thresh) & (img > thresh), 1, 0)
bmap = ndimage.maximum_filter(m, 5)
cond = (bmap == 1) & (d > thresh * sigfrac) & (img > thresh * sigfrac)
m[cond] = 1
return m
import numpy as np
import scipy.misc as sm
import scipy.ndimage as sn
import matplotlib.pyplot as mp
# 加载图像
image1 = sm.ascent().astype(np.float32)
# 均值滤波
image2 = sn.median_filter(image1, (42, 42))
# 旋转图像
image3 = sn.rotate(image1, angle=45)
# 浮雕图像
image4 = sn.prewitt(image1)
mp.gcf().set_facecolor(np.ones(3) * 240 / 255)
mp.subplot(221)
mp.title('Original', fontsize=16)
mp.xlabel('Width', fontsize=12)
mp.ylabel('Height', fontsize=12)
ax = mp.gca()
ax.xaxis.set_major_locator(mp.MultipleLocator(128))
ax.xaxis.set_minor_locator(mp.MultipleLocator(32))
ax.yaxis.set_major_locator(mp.MultipleLocator(128))
ax.yaxis.set_minor_locator(mp.MultipleLocator(32))
mp.tick_params(labelsize=10)
mp.grid(linestyle=':')
mp.imshow(image1, cmap='gray')
mp.subplot(222)
mp.title('Median', fontsize=16)
mp.xlabel('Width', fontsize=12)
mp.ylabel('Height', fontsize=12)
ax = mp.gca()
ax.xaxis.set_major_locator(mp.MultipleLocator(128))
mp.axis('off')
mp.imshow(original, cmap='gray')
# #高斯模糊
median = sn.median_filter(original, 10)
mp.subplot(2, 2, 2)
mp.axis('off')
mp.imshow(median, cmap='gray')
print(median)
#旋转
median1 = sn.rotate(original, 45)
mp.subplot(2, 2, 3)
mp.axis('off')
mp.imshow(median1, cmap='gray')
print(median1)
#边缘识别
prewitt = sn.prewitt(original)
mp.subplot(2, 2, 4)
mp.axis('off')
mp.imshow(prewitt, cmap='gray')
print(prewitt)
# #角度旋转
# rotate = sn.rotate(original, 45)
# #边缘识别
# prewitt = sn.prewitt(original)
# mp.figure('Image', facecolor='lightgray')
# mp.subplot(221)
# mp.title('Original', fontsize=16)
# mp.axis('off')
# mp.imshow(original, cmap='gray')
# mp.subplot(222)
def __init__(self,
image_matrix,
sigma=1.0,
thresHigh=40,
thresLow=6,
thresHighLimit=2**18):
self.imin = image_matrix
self.thresHigh = thresHigh
self.thresLow = thresLow
mask = numpy.ones(self.imin.shape, dtype=bool)
fsmooth = lambda x: gaussian_filter(x, sigma, mode='constant')
imout = smooth_with_function_and_mask(self.imin, fsmooth, mask)
grady = ndi.prewitt(imout, axis=1, mode='constant') * -1.0
gradx = ndi.prewitt(imout, axis=0, mode='constant')
grad = numpy.hypot(gradx, grady)
# Net gradient is the square root of sum of square of the horizontal
# and vertical gradients
# grad = numpy.hypot(gradx, grady)
theta = numpy.arctan2(grady, gradx)
theta = 180 + (180 / pi) * theta
# Only significant magnitudes are considered. All others are removed
x, y = where(grad < 10)
theta[x, y] = 0
grad[x, y] = 0
# The angles are quantized. This is the first step in non-maximum
# supression. Since, any pixel will have only 4 approach directions.
x0, y0 = where(((theta < 22.5) + (theta >= 157.5) * (theta < 202.5) +
(theta >= 337.5)) == True)
x45, y45 = where(((theta >= 22.5) * (theta < 67.5) + (theta >= 202.5) *
(theta < 247.5)) == True)
x90, y90 = where(((theta >= 67.5) * (theta < 112.5) +
(theta >= 247.5) * (theta < 292.5)) == True)
x135, y135 = where(((theta >= 112.5) * (theta < 157.5) +
(theta >= 292.5) * (theta < 337.5)) == True)
# self.theta = theta
# Image.fromarray(self.theta).convert('L').save('Angle map.jpg')
theta[x0, y0] = 0.
theta[x45, y45] = 45.
theta[x90, y90] = 90.
theta[x135, y135] = 135.
self.grad = grad[1:-1, 1:-1]
self.theta = theta[1:-1, 1:-1]
x, y = self.grad.shape
grad2 = self.grad.copy()
for i in range(x):
for j in range(y):
if self.theta[i, j] == 0.:
test = self.nms_check(grad2, i, j, 1, 0, -1, 0)
if not test:
self.grad[i, j] = 0
elif self.theta[i, j] == 45.:
test = self.nms_check(grad2, i, j, 1, -1, -1, 1)
if not test:
self.grad[i, j] = 0
elif self.theta[i, j] == 90.:
test = self.nms_check(grad2, i, j, 0, 1, 0, -1)
if not test:
self.grad[i, j] = 0
elif self.theta[i, j] == 135.:
test = self.nms_check(grad2, i, j, 1, 1, -1, -1)
if not test:
self.grad[i, j] = 0
init_point = self.initPt(thresHighLimit)
# Hysteresis tracking. Since we know that significant edges are
# continuous contours, we will exploit the same.
# thresHigh is used to track the starting point of edges and
# thresLow is used to track the whole edge till end of the edge.
self.segments = Segments.Segments()
segment = [init_point]
while init_point != -1:
# print 'next segment at',init_point
self.grad[init_point[0], init_point[1]] = -1
p2 = init_point
p1 = init_point
p0 = init_point
p0 = self.nextNbd(p0, p1, p2)
while p0 != -1:
segment.append(p0)
p2 = p1
p1 = p0
self.grad[p0[0], p0[1]] = -1
p0 = self.nextNbd(p0, p1, p2)
if len(segment) >= 2:
self.segments.append(segment)
init_point = self.nextPt(self.grad)
segment = [init_point]
self.stippleSegmentList = []
# different edge detection methods cam.capture(data, 'rgb') # capture image # diff of gaussians t0 = time.time() grad_xy = scimg.gaussian_gradient_magnitude(data[:, :, 0], sigma=1.5) ##grad_xy = np.mean(grad_xy,2) t_grad_xy = time.time() - t0 # laplacian of gaussian t0 = time.time() lap = scimg.gaussian_laplace(data[:, :, 0], sigma=0.7) t_lap = time.time() - t0 # Canny method without angle t0 = time.time() gaus = scimg.fourier_gaussian(data[:, :, 0], sigma=0.05) can_x = scimg.prewitt(gaus, axis=0) can_y = scimg.prewitt(gaus, axis=1) can = np.hypot(can_x, can_y) ##can = np.mean(can,2) t_can = time.time() - t0 # Sobel method t0 = time.time() sob_x = scimg.sobel(data[:, :, 0], axis=0) sob_y = scimg.sobel(data[:, :, 0], axis=1) sob = np.hypot(sob_x, sob_y) ##sob = np.mean(sob,2) t_sob = time.time() - t0 # plotting routines and labeling fig, ax = plt.subplots(2, 2, figsize=(12, 6)) ax[0, 0].pcolormesh(x, y, grad_xy, cmap='gray')
def eval(self, source_data):
return ndimage.prewitt(source_data, *self.args, **self.kwargs)
dataDir = "/home/arb/Delme/"
plt.imshow(griddata, origin='lower')
plt.gray()
cb = plt.colorbar()
cb.set_label('Value Range')
plt.xlabel('GridEast')
plt.ylabel('GridNorth')
plt.suptitle('Raw data')
plt.savefig(dataDir + 'r15_raw' + '.png')
plt.show()
#Calculate derivatives
gridSobel = nd.sobel(griddata)
gridLaplace = nd.laplace(griddata)
gridPrewitt = nd.prewitt(griddata)
gridGaussian = nd.gaussian_filter(griddata, 1)
gridMinimum = nd.minimum_filter(griddata, size=(3, 3))
#Plot a derivative
plt.imshow(gridGaussian, origin='lower')
plt.gray()
#show image
cb = plt.colorbar()
cb.set_label('Value Range')
plt.xlabel('GridEast')
plt.ylabel('GridNorth')
plt.suptitle('Raw data')
plt.savefig(dataDir + 'r15_gaussianDerivative' + '.png')
plt.show()
img = plt.imshow(image, cmap=plt.cm.gray)
#中值滤波器扫描信息的每一个数据点，并替换为相邻数据点的中值。对
#图像应用中值滤波器并显示在第二个子图中
plt.subplot(222)
plt.title("Median Filter")
filtered = ndimage.median_filter(image, size=(42, 42))
plt.imshow(filtered, cmap=plt.cm.gray)
#旋转图像并显示在第三个子图中
plt.subplot(223)
plt.title("Rotated")
rotated = ndimage.rotate(image, 90)
plt.imshow(rotated, cmap=plt.cm.gray)
#Prewitt滤波器是基于图像强度的梯度计算
plt.subplot(224)
plt.title("Prewitt Filter")
filtered = ndimage.prewitt(image)
plt.imshow(filtered, cmap=plt.cm.gray)
plt.show()
print u"音频处理"
#使用scipy.io.wavfile模块中的read函数可以将该文件转换为一个NumPy
#数组
#使用read函数读入文件,返回采样率和音频数据
from scipy.io import wavfile
import urllib2
import sys
sample_rate, data = wavfile.read(WAV_FILE)
#应用tile函数
repeated = np.tile(data, 4)
#使用write函数写入一个新文件
def sky_ref_patch_detection(I_origin):
"""
RETURN:
S: list:[Sb, Sg, Sr]
sky_prob_map
"""
I_gray = cv2.cvtColor(I_origin, cv2.COLOR_BGR2GRAY)
sky_prob_map = np.zeros(I_gray.shape)
# initialize sky prob map with pixels strictly within ideal blue range
sky_prob_map[ in_idea_blue_rg(I_origin) ] = 1.0
# sky is ONLY in top one third
# sky_prob_map[sky_prob_map.shape[0]*1.0/3:-1,...] = 0.0
# exponentially decrease sky prob where gradient is too large (>_grad_percent*255.)
_grad_x = np.absolute( ndi.prewitt(I_gray, axis=1 ,mode='nearest') )
_grad_y = np.absolute( ndi.prewitt(I_gray, axis=0 ,mode='nearest') )
_rv = norm(loc=0., scale=255./3.) # 3*sigma = 255, rv ~ random variable
_grad_percent = 0.10 # 0.05 of the original paper
cond_mod_sky_prob = (sky_prob_map ==1.0) & \
((_grad_x >_grad_percent*255.) | (_grad_y >_grad_percent*255.)) # average of gradx, y
sky_prob_map[ cond_mod_sky_prob ] = _rv.pdf( (_grad_x + _grad_y)/2.0 )[ cond_mod_sky_prob ]
# detect bimodal
_L = cv2.cvtColor(I_origin, cv2.COLOR_BGR2LAB)[...,0]
detect_res = cape_util.detect_bimodal(
[cape_util.get_smoothed_hist( cape_util.mask_skin(_L, sky_prob_map!=0.0) )]
)[0] # detect_bimodal is an array f
if (detect_res[0] == True): #could return F or None, must be ==True
print 'bimodal detected in current sky_ref_patch'
sky_prob_map[ I_gray ==detect_res[1] ] = 0.0 #exclude pixels correspond to the dark mode
# get mean and std from each b,g,r channel of detected sky
S = []
if ( np.sum( _get_top_one_third(sky_prob_map) ) !=0): # top 1/3 has sky
sky_prob_map_bgr = _2to3(sky_prob_map)
for i in range(3): #B, G, R
masked_I_one_thrid = cape_util.mask_skin( # mask non-blue area 0, copy the rest
_get_top_one_third(I_origin)[...,i]
, _get_top_one_third(sky_prob_map)!=0.0 # sky_prob map changed after previous step
)
mean, std = get_sky_ref_patch( masked_I_one_thrid, sky_prob_map )
S.append(mean)
_rv_sky_patch = norm(loc=mean, scale=std)
# re-assign (where sky prob>0), normalize to p(median) = 1.0
sky_prob_map_bgr[...,i][sky_prob_map>0.0] = \
_rv_sky_patch.pdf(I_origin[...,i])[sky_prob_map>0.0] / _rv_sky_patch.pdf(mean)
_b=1.; _g=5.; _r=3.
sky_prob_map = (_b*sky_prob_map_bgr[...,0] + _g*sky_prob_map_bgr[...,1] + _r*sky_prob_map_bgr[...,2]) / (_b+_g+_r)
# remove small patches: image opening on sky_prob_map
_h,_w = sky_prob_map.shape[0:2]
_kernel = cv2.getStructuringElement(cv2.MORPH_ELLIPSE,(_h/12,_w/16))
sky_mask_opened = cv2.morphologyEx(sky_prob_to_01(sky_prob_map).astype('uint8'), cv2.MORPH_OPEN, _kernel)
sky_prob_map[sky_mask_opened==0.0] = 0.0 # set sky_prob_map to 0 where removed from sky_mask_opened
else:
print 'sky not detected in top 1/3'
plt.imshow(sky_prob_map); plt.show() # rainbow map
print 'S(b,g,r): ',S
return S, sky_prob_map
def run(self, ips, img, buf, para=None):
nimg.prewitt(img, output=buf)
import numpy as np
import matplotlib.pyplot as plt
from scipy import ndimage
image = misc.lena().astype(np.float32)
plt.subplot(221)
plt.title("Original Image")
img = plt.imshow(image, cmap=plt.cm.gray)
plt.axis("off")
plt.subplot(222)
plt.title("Median Filter")
filtered = ndimage.median_filter(image, size=(42,42))
plt.imshow(filtered, cmap=plt.cm.gray)
plt.axis("off")
plt.subplot(223)
plt.title("Rotated")
rotated = ndimage.rotate(image, 90)
plt.imshow(rotated, cmap=plt.cm.gray)
plt.axis("off")
plt.subplot(224)
plt.title("Prewitt Filter")
filtered = ndimage.prewitt(image)
plt.imshow(filtered, cmap=plt.cm.gray)
plt.axis("off")
plt.show()
def prewitt_image(image):
return sn.prewitt(image)
from sklearn.naive_bayes import GaussianNB
from sklearn.ensemble import BaggingClassifier
from scipy import ndimage
from sklearn import linear_model
import matplotlib.pyplot as plt
Targets = np.genfromtxt("data/targets.csv")
Data = []
for i in range(1, 279):
imagefile = nib.load("data/set_train/train_" + str(i) + ".nii")
image = imagefile.get_data()
I = image[:, :, :, 0]
I = np.asarray(I, dtype='float')
I = ndimage.prewitt(I, axis=0)
I = ndimage.gaussian_filter(I, sigma=1)
imagefile.uncache()
Data.append(np.asarray(I))
Data = np.asarray(Data)
X_train, X_test, y_train, y_test = \
train_test_split(Data, Targets, test_size=0.33, random_state=42, stratify=Targets)
#X_train = Data
#y_train = Targets
print "fitting has started"
clf = linear_model.LogisticRegression()
def reliefshade(deminfo, dem, surface):
vertical_exaggeration = 2.0
# light_gamma < 1 brings out detail in the flats, hides it in hills facing the lightsource
light_gamma = 0.9
light_level = 0.5
# shade_gamma > 1 brings out detail in darkest slopes at the expense of some gentle ones
shade_gamma = 2.0
shade_level = 0.5
# imhof sez the light is golden - #fffbf4 maybe?
# and the shade is purple - #060009 maybe?
#lightcolor = (1.0, 0.98, 0.95)
#shadecolor = (1.0, 0.98, 0.95)
#shadecolor = (0.7, 0.8, 1.0)
#lightcolor = (1.0, 1.0, 1.0)
lightcolor = (1.0, 1.0, 0.9)
shadecolor = (1.0, 1.0, 1.0)
#shadecolor = (0.8, 0.8, 1.0)
# scale adjusts for horizontal and vertical units as well as
# any height exaggeration.
scale = vertical_exaggeration / deminfo.grid.meters_per_grid()
# 2d prewitt filter kernel with axis=1 is [[-1, 0, 1]
# [-1, 0, 1]
# [-1, 0, 1]]
dvdx = scale * ndimage.prewitt(dem, axis=1, mode='nearest')
dvdy = -scale * ndimage.prewitt(dem, axis=0, mode='nearest')
#glumpy_loop(N.outer(dvdx, (1,1,1)).reshape(surface.shape))
# the idea of separating light and shadow is from lars
# ahlzen's toposm work, this is slightly different.
# normal vector is (-dvdx, -dvdy, 1)
# norm of normal is sqrt(dvdx * dvdx + dvdy * dvdy + 1)
# vector toward light is (-1, 1, 1)
# norm of light vector sqrt(3)
norm = N.sqrt(3 * (dvdx * dvdx + dvdy * dvdy + 1))
# light_dir = (-1, 1, 1)
# cos(theta) = light_dir . normal / |normal| |light_dir|
#
costh = (dvdx - dvdy + 1) / norm
# separate into illumination and (self-)shadow
# costh_level is costh for a horizontal surface. anything brighter
# than this is light, anything darker is shade.
# this is (light=0,shade=1)
costh_level = 1.0 / sqrt(3)
# the minimum possible costh is for a cliff facing southwest
# which would have dvdx=-BIG and dvdy=+BIG. plug that into
# the costh formula and you get -sqrt(2/3)
# this is the darkest possible shade (light=0,shade=0)
costh_min = -sqrt(2.0 / 3.0)
# light is 1.0 for a surface pointing sunward and 0.0 at horizontal
light = N.clip((costh - costh_level) / (1 - costh_level), 0, 1)
light = light ** (1/light_gamma)
# shade is 1.0 for horizontal and 0.0 for a vertical cliff facing
# away from the lightsource.
# note that shade=0 is still darkest, 1 is lightest
shade = N.clip((costh-costh_min) / (costh_level-costh_min), 0, 1)
shade = shade ** (1/shade_gamma)
# this would be a great place for a "tone mapping" like step
# to adapt the relief to meadows or mountains.
# N.outer flattens so must reshape to rgb afterward
light = N.outer(light, lightcolor).reshape(surface.shape)
shade = N.outer(shade, shadecolor).reshape(surface.shape)
# moderate the effect of light and shade here
light = light * light_level
shade = shade * shade_level + (1 - shade_level)
# light brings the surface closer to the light color
# XXX the white highlights look like plastic, mix in
# the surface color a bit more?
lit = (1 - (1-surface) * (1-light))
# shade darkens to black anywhere
relief = lit * shade
return relief
