def main():
figa = np.array(Image.open('images/Fig1130(a)(uniform_noise).tif').convert('L'))
figb = np.array(Image.open('images/Fig1130(b)(sinusoidal).tif').convert('L'))
figc = np.array(Image.open('images/Fig1130(c)(cktboard_section).tif').convert('L'))
glcma = glcm(figa, (0, 1))
glcmb = glcm(figb, (0, 1))
glcmc = glcm(figc, (0, 1))
fa = glcm_features(glcma)
fb = glcm_features(glcmb)
fc = glcm_features(glcmc)
ax = plt.subplot('321')
ax.set_title("Figure a")
ax.imshow(figa, 'gray')
ax.axis('off')
ax = plt.subplot('322')
ax.set_title("GLCM of Figure b")
ax.imshow(adjust_log(glcma), 'gray')
ax.axis('off')
ax = plt.subplot('323')
ax.set_title("Figure b")
ax.imshow(figb, 'gray')
ax.axis('off')
ax = plt.subplot('324')
ax.set_title("GLCM of Figure b")
ax.imshow(adjust_log(glcmb), 'gray')
ax.axis('off')
ax = plt.subplot('325')
ax.set_title("Figure c")
ax.imshow(figc, 'gray')
ax.axis('off')
ax = plt.subplot('326')
ax.set_title("GLCM of Figure c")
plt.imshow(adjust_log(glcmc), 'gray')
ax.axis('off')
plt.show()
def show_sarimage3d(SI,
pdict,
axismod=None,
title=None,
cmap=None,
isimgadj=False,
aspect=None,
outfile=None,
figsize=None):
r"""[summary]
[description]
Arguments:
SI {[type]} -- [description]
sarplat {[type]} -- [description]
Keyword Arguments:
axismod {[type]} -- [description] (default: {None})
title {[type]} -- [description] (default: {None})
cmap {[type]} -- [description] (default: {None})
isimgadj bool -- [description] (default: {False})
aspect {[type]} -- [description] (default: {None})
outfile {[type]} -- [description] (default: {None})
figsize {[type]} -- [description] (default: {None})
"""
extent, xlabelstr, ylabelstr = saraxis(pdict=pdict, axismod=axismod)
Z = np.absolute(SI)
if isimgadj:
Z = exposure.adjust_log(Z)
Z = np.flipud(Z)
# fig = plt.figure(figsize=figsize)
M, N = np.shape(Z)
X = np.arange(0, N, 1)
Y = np.arange(0, M, 1)
X, Y = np.meshgrid(X, Y)
# mlab.figure(size=(400, 500))
# mlab.mesh(X, Y, Z)
# mlab.surf(X, Y, Z)
# mlab.colorbar()
# mlab.xlabel(xlabelstr)
# mlab.ylabel(ylabelstr)
# mlab.zlabel("Amplitude")
# mlab.title(title)
# mlab.show()
if outfile is not None:
# mlab.savefig(outfile)
print("sar image has been saved to: ", outfile)
def open_tiff_files(self):
self.tiff_file_to_view = QtWidgets.QFileDialog.getOpenFileName(
directory=self.user_dir, filter='*.tiff', parent=self)[0]
img = plt.imread(self.tiff_file_to_view)
logarithmic_corrected = exposure.adjust_log(img, 1)
self.figureiff_image.ax.clear()
self.figure_tiff_image.ax.imshow(logarithmic_corrected,
cmap='BuPu_r',
vmax=2000)
self.canvas_tiff_image.draw_idle()
def log_compression(file_name, img):
"""
Replace pixel value with its logarithm (effectively enhancing low intensity
pixel values).
:param file_name: file name of the image as used to save the image.
:param img: the jpeg image.
:return: output image (in .jpg)
"""
log_out = exposure.adjust_log(img)
log_hist(file_name, log_out)
img_log = io.imsave(file_name + '_log.jpg', log_out)
return log_out
def logcorr(self):
"""logarithmic exposure added to image
Parameters
----------
image : n-dimensional array
Input image folder.
Returns
-------
Image with logarathmic exposure.
"""
return exposure.adjust_log(self)
def log_compression(img):
"""Performs log compression processing on raw image
Args:
img (np array): raw image in the form of a np array
Returns:
np array: image array after having log compression
performed
"""
img_log = exposure.adjust_log(img, 1)
logging.info('Log compression performed!')
return img_log
def log_correct_img(img):
"""
Take address of image to be processed as img_loc.
Computes log correction and returns both the input img and corrected img.
param:
img - address to access image to be processed
returns:
img - 2D array of input image
log_img - 2D array of log corrected image
"""
log_img = np.asarray(adjust_log(img, 2), dtype='uint8')
return log_img[:, :, 0:3]
def createTFRecord(filename, mapfile, num):
class_map = {}
data_dir = 'dataset/'
data_dir_d = 'dataset_d/'
classes = {
'apple', 'ball', 'banana', 'bowl', 'garlic', 'green', 'lemon',
'mushroom', 'onion', 'orange', 'peach', 'pear', 'potato', 'tomato'
}
# 输出TFRecord文件的地址
writer = tf.python_io.TFRecordWriter(filename)
i = 0
for index, name in enumerate(classes):
class_path = data_dir + name + '/'
class_map[index] = name
for img_name in os.listdir(class_path):
i += 1
if (i % num == 0) and (i % (num * 10) != 0):
img_path = class_path + img_name # 每个图片的地址
img_path_d = data_dir_d + name + '/' + img_name
img = Image.open(img_path)
img = img.resize((208, 208))
img = img.convert("RGB")
img_raw = img.tobytes() # 将图片转化
# 成二进制格式
img_d = Image.open(img_path_d)
img_d = img_d.resize((208, 208))
img_d = img_d.convert("I")
img_d = img_as_float(img_d)
img_d = exposure.adjust_gamma(img_d, 1.0)
img_d = exposure.adjust_log(img_d, 100000000)
img_d = exposure.rescale_intensity(img_d,
in_range='image',
out_range=np.uint8)
img_raw_d = img_d.tobytes()
example = tf.train.Example(features=tf.train.Features(
feature={
'label': _int64_feature(index),
'image_raw': _bytes_feature(img_raw),
'image_raw_d': _bytes_feature(img_raw_d)
}))
writer.write(example.SerializeToString())
writer.close()
txtfile = open(mapfile, 'w+')
for key in class_map.keys():
txtfile.writelines(str(key) + ":" + class_map[key] + "\n")
txtfile.close()
def test_adjust_log():
"""Verifying the output with expected results for logarithmic
correction with multiplier constant multiplier equal to unity"""
image = np.arange(0, 255, 4, np.uint8).reshape(8,8)
expected = np.array([[ 0, 5, 11, 16, 22, 27, 33, 38],
[ 43, 48, 53, 58, 63, 68, 73, 77],
[ 82, 86, 91, 95, 100, 104, 109, 113],
[117, 121, 125, 129, 133, 137, 141, 145],
[149, 153, 157, 160, 164, 168, 172, 175],
[179, 182, 186, 189, 193, 196, 199, 203],
[206, 209, 213, 216, 219, 222, 225, 228],
[231, 234, 238, 241, 244, 246, 249, 252]], dtype=np.uint8)
result = exposure.adjust_log(image, 1)
assert_array_equal(result, expected)
def test_adjust_inv_log():
"""Verifying the output with expected results for inverse logarithmic
correction with multiplier constant multiplier equal to unity"""
image = np.arange(0, 255, 4, np.uint8).reshape(8,8)
expected = np.array([[ 0, 2, 5, 8, 11, 14, 17, 20],
[ 23, 26, 29, 32, 35, 38, 41, 45],
[ 48, 51, 55, 58, 61, 65, 68, 72],
[ 76, 79, 83, 87, 90, 94, 98, 102],
[106, 110, 114, 118, 122, 126, 130, 134],
[138, 143, 147, 151, 156, 160, 165, 170],
[174, 179, 184, 188, 193, 198, 203, 208],
[213, 218, 224, 229, 234, 239, 245, 250]], dtype=np.uint8)
result = exposure.adjust_log(image, 1, True)
assert_array_equal(result, expected)
def augmentations(image_array: ndarray):
v_min, v_max = np.percentile(image_array, (0.2, 99.8))
return (
image_array,
transform.rotate(image_array, random.uniform(-50, 50)),
exposure.rescale_intensity(image_array, in_range=(v_min, v_max)),
util.random_noise(image_array),
ndimage.gaussian_filter(image_array, 2),
exposure.adjust_log(image_array),
exposure.adjust_sigmoid(image_array),
#color.rgb2gray(image_array), (FOR COLORED IMAGES)
#np.invert(image_array), (FOR COLORED IMAGES)
exposure.adjust_gamma(image_array, gamma=0.4, gain=0.9),
image_array[:, ::-1],
image_array[::-1, :])
def adjust(frame, method, **kwargs):
if method == "equalize":
adjusted = exp.equalize_hist(frame, **kwargs)
elif method == "gamma":
adjusted = exp.adjust_gamma(frame, **kwargs)
elif method == "log":
adjusted = exp.adjust_log(frame, **kwargs)
elif method == "sigmoid":
adjusted = exp.adjust_sigmoid(frame, **kwargs)
elif method == "adaptive":
adjusted = exp.equalize_adapthist(frame, **kwargs)
else:
raise ValueError(
"method can be equalize, gamma, log, sigmoid or adaptive")
return adjusted
def test_adjust_inv_log():
"""Verifying the output with expected results for inverse logarithmic
correction with multiplier constant multiplier equal to unity"""
image = np.arange(0, 255, 4, np.uint8).reshape((8, 8))
expected = np.array(
[[0, 2, 5, 8, 11, 14, 17, 20], [23, 26, 29, 32, 35, 38, 41, 45],
[48, 51, 55, 58, 61, 65, 68, 72], [76, 79, 83, 87, 90, 94, 98, 102],
[106, 110, 114, 118, 122, 126, 130, 134],
[138, 143, 147, 151, 156, 160, 165, 170],
[174, 179, 184, 188, 193, 198, 203, 208],
[213, 218, 224, 229, 234, 239, 245, 250]],
dtype=np.uint8)
result = exposure.adjust_log(image, 1, True)
assert_array_equal(result, expected)
def detect_watershed_patches(A,
mask=0,
validation=False,
blobs=None,
label=''):
A_copy = np.copy(A)
A_copy[A_copy < mask] = 0
A_log = adjust_log(A_copy)
A_norm = rescale_intensity(A_log)
print('Raster shape (%d,%d)' % A_norm.shape)
if blobs is None:
blobs = blob_dog(A_norm, max_sigma=30, threshold=.1)
print('Detected %d blobs to seed watershed' % len(blobs))
else:
print('Seeding watershed with {len} predefined blobs'.format(
len=len(blobs)))
if validation:
title = 'DetectBlobs' if label == '' else '{label} Blobs'.format(
label=label)
ax = plot(A_copy, title)
for blob in blobs:
y, x, sigma = blob
# plot blobs as circles. NOTE: multiply 'r' by sqrt(2) because the blob_dog docs say that the
# radius of each blob is approximately sqrt(2)*(std. deviation of that blob's Gaussian kernel)
c = plt.Circle((x, y),
sqrt(2) * sigma,
color='k',
linewidth=1,
fill=False)
ax.add_patch(c)
x, y = np.indices(A_norm.shape)
markers = np.zeros(A_norm.shape)
idxs = range(len(blobs))
random.shuffle(idxs)
for i, blob in zip(idxs, blobs):
xb, yb, rb = blob
mask_circle = (x - xb)**2 + (y - yb)**2 < rb**2
markers[mask_circle] = i + 1
watershed_patches = watershed(1 - A_norm, markers, mask=A_copy)
if validation:
title = 'Watershed' if label == '' else 'Watershed {label}'.format(
label=label)
plt.figure(title, figsize=(8, 8))
watershed_patches[watershed_patches < 1] = np.nan
plt.imshow(watershed_patches, cmap='Paired')
plt.tight_layout()
plt.title(title)
return watershed_patches, len(blobs)
def test_adjust_log():
"""Verifying the output with expected results for logarithmic
correction with multiplier constant multiplier equal to unity"""
image = np.arange(0, 255, 4, np.uint8).reshape((8, 8))
expected = np.array(
[[0, 5, 11, 16, 22, 27, 33, 38], [43, 48, 53, 58, 63, 68, 73, 77],
[82, 86, 91, 95, 100, 104, 109, 113],
[117, 121, 125, 129, 133, 137, 141, 145],
[149, 153, 157, 160, 164, 168, 172, 175],
[179, 182, 186, 189, 193, 196, 199, 203],
[206, 209, 213, 216, 219, 222, 225, 228],
[231, 234, 238, 241, 244, 246, 249, 252]],
dtype=np.uint8)
result = exposure.adjust_log(image, 1)
assert_array_equal(result, expected)
def receive_blue_images(imgs):
blue_imgs = []
for img in imgs:
h, w, t = img.shape
temp = img.copy()
temp = exposure.adjust_log(temp)
mean = int(round(np.mean(temp)) * 1.1)
for i in range(h):
for j in range(w):
if (temp[i][j][2] < temp[i][j][1]) or (temp[i][j][2] <
temp[i][j][0]):
temp[i][j][0] = mean
temp[i][j][1] = mean
temp[i][j][2] = mean
blue_imgs.append(temp)
return blue_imgs
def adjust_log(img):
image = exposure.adjust_log(img) #对数调整
plt.figure('adjust_gamma', figsize=(8, 8))
plt.subplot(121)
plt.title('origin image')
plt.imshow(img, plt.cm.gray)
plt.axis('off')
plt.subplot(122)
plt.title('log')
plt.imshow(image, plt.cm.gray)
plt.axis('off')
plt.show()
return face_detection(image)
def image_transformation(X, method_type='blur', **kwargs):
# https://www.kaggle.com/tomahim/image-manipulation-augmentation-with-skimage
q = kwargs['percentile'] if 'percentile' in kwargs else (0.2, 99.8)
angle = kwargs['angle'] if 'angle' in kwargs else 60
transformation_dict = {
'blur': normalize(ndimage.uniform_filter(X)),
'invert': normalize(util.invert(X)),
'rotate': rotate(X, angle=angle),
'rescale_intensity': _rescale_intensity(X, q=q),
'gamma_correction': exposure.adjust_gamma(X, gamma=0.4, gain=0.9),
'log_correction': exposure.adjust_log(X),
'sigmoid_correction': exposure.adjust_sigmoid(X),
'horizontal_flip': X[:, ::-1],
'vertical_flip': X[::-1, :],
'rgb2gray': skimage.color.rgb2gray(X)
}
return transformation_dict[method_type]
def _rgb(img, band=('RED','GREEN','BLUE')):
gain = 3
scale = 1E4
#READ BANDS
red = img.feature(band[0], dtype=np.float32)
green = img.feature(band[1], dtype=np.float32)
blue = img.feature(band[2], dtype=np.float32)
#PROCESS RGB
RGB = np.stack( ( red, green, blue ), axis=2 )
RGB = RGB/scale #normalize float values to [0;1]
RGB = adjust_log(RGB,gain) # adjust gamma
RGB[RGB>1] = 1 # clip saturated values
RGB[RGB<0] = 0
return RGB
def image_transformation(X, method_type='blur', **kwargs):
# https://www.kaggle.com/tomahim/image-manipulation-augmentation-with-skimage
q = kwargs['percentile'] if 'percentile' in kwargs else (0.2, 99.8)
angle = kwargs['angle'] if 'angle' in kwargs else 60
transformation_dict = {
'blur': normalize(ndimage.uniform_filter(X)),
'invert': normalize(util.invert(X)),
'rotate': rotate(X, angle=angle),
'rescale_intensity': _rescale_intensity(X, q=q),
'gamma_correction': exposure.adjust_gamma(X, gamma=0.4, gain=0.9),
'log_correction': exposure.adjust_log(X),
'sigmoid_correction': exposure.adjust_sigmoid(X),
'horizontal_flip': X[:, ::-1],
'vertical_flip': X[::-1, :],
'rgb2gray': skimage.color.rgb2gray(X)
}
return transformation_dict[method_type]
def m5():
Image.MAX_IMAGE_PIXELS = None
for fn in os.listdir("all"):
if fn.endswith("TIF"):
adj_fn = fn.split("_B")[1]
if len(adj_fn) == 5:
adj_fn = "0" + adj_fn
adj_fn = "A_" + adj_fn[:-4]
print(adj_fn)
image = io.imread("all/" + fn)
print(image.shape)
if fn.endswith("8.TIF"):
pass
region = image[3500 * 2:3900 * 2, 800 * 2:1200 * 2]
else:
region = image[3500:3900, 800:1200]
region = transform.resize(region, (800, 800),
anti_aliasing=False)
io.imsave("edited/" + adj_fn + ".png", region)
exp = exposure.equalize_adapthist(region)
io.imsave("edited/" + adj_fn + "adj_adaphist.png", exp)
region = exposure.equalize_hist(region)
io.imsave("edited/" + adj_fn + "adj_hist.png", region)
exp = exposure.adjust_sigmoid(region)
io.imsave("edited/" + adj_fn + "adj_sig.png", exp)
exp = exposure.adjust_gamma(region)
io.imsave("edited/" + adj_fn + "adj_gam.png", exp)
exp = exposure.adjust_log(region)
io.imsave("edited/" + adj_fn + "adj_log.png", exp)
# exp = exposure.equalize_hist(region)
# io.imsave("edited/"+adj_fn +"adj_hist.png", exp)
#roberts, sobel, scharr, prewitt
edges = filters.roberts(exp)
io.imsave("edited/" + adj_fn + "f_rob.png", edges)
edges = filters.sobel(exp)
io.imsave("edited/" + adj_fn + "f_sob.png", edges)
def create_mask_pass1(img):
"""
This is the first pass in creating a mask. The skimage.exposure is used as it helps to get rid
of the glue around the tissue. This also calls the compute_mask method which is part
of the nipy package which you will need to install from github
:param img: the raw image
:return: the 1st pass of the image
"""
img = exposure.adjust_log(img, 1)
img = exposure.adjust_gamma(img, 2)
mask = compute_mask(img, m=0.2, M=0.9, cc=False, opening=2, exclude_zeros=True)
mask = mask.astype(int)
mask[mask==0] = 0
mask[mask==1] = 255
kernel = np.ones((5, 5), np.uint8)
mask = cv2.dilate(mask.astype(np.uint8), kernel, iterations=2)
mask = mask.astype(np.uint8)
return mask
def standardize(self, x):
if self.rescale:
x *= self.rescale
# x is a single image, so it doesn't have image number at index 0
img_channel_index = self.channel_index - 1
grayscale = x.shape[img_channel_index] == 1
if self.samplewise_center:
if grayscale:
raise ValueError(
'samplewise_center on a grey image does not make sense')
x -= np.mean(x, axis=img_channel_index, keepdims=True)
if self.samplewise_std_normalization:
x /= (np.std(x, axis=img_channel_index, keepdims=True) + 1e-7)
if self.featurewise_center:
x -= self.mean
if self.featurewise_std_normalization:
x /= (self.std + 1e-7)
if self.zca_whitening:
flatx = np.reshape(x, (x.size))
whitex = np.dot(flatx, self.principal_components)
x = np.reshape(whitex, (x.shape[0], x.shape[1], x.shape[2]))
# Add our custome stuff
# vgg prerpocessing
if self.bmodel_preprocessing is not None:
module = getattr(bmodel, self.bmodel_preprocessing.lower())
s = x.shape
x = np.expand_dims(x, axis=0)
x = getattr(module, 'preprocess_input')(x)
x = x.reshape(s)
# add_channel to gray images
if x.shape[img_channel_index] == 1:
x = x.transpose(2, 0, 1)
if self.add_channel:
log = exposure.adjust_log(x)
hist = exposure.equalize_adapthist(
x[0].astype('int16'))[np.newaxis]
x = np.vstack((log, hist, x)).transpose(1, 2, 0)
else:
x = np.vstack(([x] * 3)).transpose(1, 2, 0)
return x
def process_image(image, list_processing_method, actions):
"""
Function process_image allows user to conduct the following
image processing methods.
Author: Haitong Wang
Date: Dec, 7th, 2018
Version: 1.0.0
:param image: 2d grayscale array
:param list_processing_method:
:param actions: array of integer indicates the number of
processing actions
:return: output image/processed image, actions, size of
the image
"""
output = image
for n in list_processing_method:
if n is "HE":
output = exposure.equalize_hist(output)
output = exposure.rescale_intensity(output, out_range=(0, 255))
actions[0] += 1
elif n is "CS":
p5, p95 = np.percentile(output, (5, 95))
output = exposure.rescale_intensity(output, in_range=(p5, p95))
actions[1] += 1
elif n is "LC":
output = exposure.adjust_log(output, 1)
output = exposure.rescale_intensity(output, out_range=(0, 255))
actions[2] += 1
elif n is "RV":
output = util.invert(output)
output = exposure.rescale_intensity(output, out_range=(0, 255))
actions[3] += 1
elif n is "GC":
output = exposure.adjust_gamma(output, 2)
output = exposure.rescale_intensity(output, out_range=(0, 255))
actions[4] += 1
size = image.shape
# add another feature for the phase 2
return output, actions, size
def image_augmentation(original_image, i, j):
aug_list = []
image_rescaled = rescale(original_image, 1.0 / 4.0)
cv2.imwrite(path2 + str(i) + str(j) + ".jpg", image_rescaled)
image_with_random_noise = random_noise(original_image)
cv2.imwrite(path2 + str(i) + str(j) + ".jpg", image_with_random_noise)
gray_scale_image = rgb2gray(original_image)
aug_list.append(gray_scale_image)
color_inversion_image = util.invert(original_image)
aug_list.append(color_inversion_image)
image_with_rotation = rotate(original_image, 45)
aug_list.append(image_with_rotation)
v_min, v_max = np.percentile(original_image, (0.2, 99.8))
better_contrast = exposure.rescale_intensity(original_image,
in_range=(v_min, v_max))
aug_list.append(better_contrast)
#adjusted_gamma_image = exposure.adjust_gamma(original_image, gamma=0.4, gain=0.9)
#aug_list.append(adjust_gamma)
log_correction_image = exposure.adjust_log(original_image)
aug_list.append(log_correction_image)
sigmoid_correction_image = exposure.adjust_sigmoid(original_image)
aug_list.append(sigmoid_correction_image)
horizontal_flip = original_image[:, ::-1]
aug_list.append(horizontal_flip)
vertical_flip = original_image[::-1, :]
aug_list.append(vertical_flip)
blured_image = ndimage.uniform_filter(original_image, size=(11, 11, 1))
aug_list.append(blured_image)
return aug_list
def main(image):
img = image
# Gamma
gamma_corrected = exposure.adjust_gamma(img, 2)
# Logarithmic
logarithmic_corrected = exposure.adjust_log(img, 1)
# Display results
fig = plt.figure(figsize=(8, 5))
axes = np.zeros((2, 3), dtype=np.object)
axes[0, 0] = plt.subplot(2, 3, 1)
axes[0, 1] = plt.subplot(2, 3, 2, sharex=axes[0, 0], sharey=axes[0, 0])
axes[0, 2] = plt.subplot(2, 3, 3, sharex=axes[0, 0], sharey=axes[0, 0])
axes[1, 0] = plt.subplot(2, 3, 4)
axes[1, 1] = plt.subplot(2, 3, 5)
axes[1, 2] = plt.subplot(2, 3, 6)
ax_img, ax_hist, ax_cdf = plot_img_and_hist(img, axes[:, 0])
ax_img.set_title('Low contrast image')
y_min, y_max = ax_hist.get_ylim()
ax_hist.set_ylabel('Number of pixels')
ax_hist.set_yticks(np.linspace(0, y_max, 5))
ax_img, ax_hist, ax_cdf = plot_img_and_hist(gamma_corrected, axes[:, 1])
ax_img.set_title('Gamma correction')
ax_img, ax_hist, ax_cdf = plot_img_and_hist(logarithmic_corrected, axes[:,
2])
ax_img.set_title('Logarithmic correction')
ax_cdf.set_ylabel('Fraction of total intensity')
ax_cdf.set_yticks(np.linspace(0, 1, 5))
# prevent overlap of y-axis labels
fig.tight_layout()
return plt, gamma_corrected, logarithmic_corrected
def getmask(rgb, fmedian=True):
"""Get eye mask"""
if (len(rgb.shape) > 3):
rgb = rgb[0, ...]
# Convert to Grayscale
gray = rgb2gray(rgb)
# Adjust Contrast
gray = gray - np.min(gray)
gray = gray / np.max(gray)
# Contrast Enhancement
gray = adjust_log(gray, 5)
# Apply median filter to remove outliers
if fmedian:
gray = medfilt(gray, 5)
# Threshold
mask = np.zeros(gray.shape)
mask[np.where(gray > np.min(gray))] = 1
mask = np.asarray(mask)
return mask
def Contrast(filein, fileout):
image = cv2.imread(filein, 0)
gam1 = exposure.adjust_log(image) # 对数调整
count = 1
for i in range(8, 15):
if i == 10:
break
gam = exposure.adjust_gamma(image, 0.1 * i)
cv2.imwrite(fileout + '_con' + str(count) + '.tif', gam)
count += 1
for i in range(4, 6):
for j in range(6, 10):
if i == 5 and j == 5:
break
img = np.array(image)
mean = np.mean(img)
img = img - mean
img = img * 0.2 * j + mean * 0.2 * i # 修对比度和亮度
cv2.imwrite(fileout + '_con' + str(count) + '.tif', img)
count += 1
def sPipe(image):
print "Reading Image"
t0 = time.time()
retina = readImage(image)
t1 = time.time()
print("Done took %.2f seconds" % (t1 - t0))
print 'Converting to Grayscale with luminance preservation'
t0 = time.time()
retina_gray = convertImage(retina)
t1 = time.time()
print("Done took %.2f seconds" % (t1 - t0))
print 'Cropping image from background'
t0 = time.time()
retina_crop = cropImage(retina_gray, 0.05)
t1 = time.time()
print("Done took %.2f seconds" % (t1 - t0))
print 'Logarithmic Correction'
t0 = time.time()
retina_adjust = exposure.adjust_log(retina_crop)
t1 = time.time()
print("Done took %.2f seconds" % (t1 - t0))
viewImage(retina_adjust)
return retina_adjust
def process_imgs_with_method(list_of_decoded_imgs, procedure):
"""Performs the image processing procedure on a list of images
Args:
list_of_decoded_imgs (list): all images from the request in matrix form
procedure (str): specifies which procedure to perform
Returns:
list: all images after applying the procedure
"""
list_of_processed_imgs = []
for before_filtering in list_of_decoded_imgs:
if procedure == 'histogram_eq':
processed = exposure.equalize_hist(before_filtering)
elif procedure == 'contrast_str':
p2, p98 = np.percentile(before_filtering, (2, 98))
processed = exposure.rescale_intensity(before_filtering,
in_range=(p2, p98))
elif procedure == 'log_compress':
processed = exposure.adjust_log(before_filtering)
else:
processed = invert(before_filtering)
list_of_processed_imgs.append(processed)
return list_of_processed_imgs
def zbar_decode(img):
from pyzbar.pyzbar import decode as pyzbar_decode
#img = img.mean(axis=2)
res=pyzbar_decode(img)
if len(res)==0:
if len(img.shape)>2:
imgg = img.mean(axis=2)
res=pyzbar_decode(imgg)
if len(res)==0:
imgb = gaussian(img,1,preserve_range=True)
res=pyzbar_decode(imgb)
if len(res)==0:
imgc = adjust_log(img)
res=pyzbar_decode(imgc)
#if len(res)==0:
# imgs = unsharp_mask(img,preserve_range=True)
# res=pyzbar_decode(imgs)
if len(res)==0:
return None
elif len(res)==1:
return res[0].data.decode("utf-8")
else:
return [r.data.decode("utf-8") for r in res]
def area(x, y):
AREA = label(x, background=0)
#print(AREA)
plt.imsave(str(m) + y + 'Area.tif', numpy.array(AREA))
properties = regionprops(AREA)
proparea = [p.area for p in properties]
#centre mean of image
proparea2 = [p.centroid for p in properties]
#co-ordinates
proparea3 = [p.coords for p in properties]
areatype = img_as_ubyte(x)
areatype = grey2rgb(areatype)
areatype = adjust_log(areatype, gain=50)
for a in proparea2:
#print(a[0])
#print(a[1])
rr, cc = circle(a[0], a[1], 1)
areatype[rr, cc] = (0, 255, 255)
proparea = DataFrame(proparea, columns=['Area'])
proparea2 = DataFrame(proparea2, columns=['row - y', 'column - x'])
merged = concat([proparea, proparea2], axis=1)
plt.imsave(str(m) + y + '_arearegions.tif', numpy.array(areatype))
merged.to_csv(str(m) + y + '_proparea.csv')
print("Calculated area saved!")
def gen_log(image):
images_log, images_otsu, images_medfilt = [], [], []
start_log = 0.006
step_log = 0.001
for i in range(11):
p = start_log + i * step_log
im_log = ex.adjust_log(image, p)
thresh = threshold_otsu(im_log)
im_log_otsu = (im_log > thresh) * 1
im_log_medfilt = medfilt(im_log)
im_log = normalize(im_log)
im_log_otsu = normalize(im_log_otsu)
im_log_medfilt = normalize(im_log_medfilt)
images_log.append(im_log.astype(int))
images_otsu.append(im_log_otsu.astype(int))
images_medfilt.append(im_log_medfilt.astype(int))
return images_log, images_otsu, images_medfilt
def extract_cbf(cbf_img, seed, brain_mask):
'''
Input:
1.the cbf image getting from Ais-system
2.seed is the location of maximal lesion value in irf image
3.brain_mask is the main matter of brain
Output:
lesion core extracted from cbf image
Describe:
1. Enhence the contrast ratio of Input Image using Clahe.
2. Calculate the power image at x and y directions, and conbine together.
3. Enhencing the contrast ratio by expanding distribution between [0, 1] using log correction function, a * log(1 + x).
4. Split the power image into two parts, basin and ridge, using otsu or cluster.
5. Locating the target basin area through seed location, the ridge around target basin is contour of lesion in cbf image.
6. Mapping pixel location into distance from seed location.
7. Spliting brain into three parts, drop off the part farthest away from target area and select the part closest to seed location.
'''
h, w = cbf_img.shape
x, y = seed
aeh_img = exposure.equalize_adapthist(cbf_img / 255, kernel_size=None, clip_limit=0.01, nbins=256)
sobel_x = cv2.Sobel(aeh_img, cv2.CV_64F, 1, 0, ksize=3)
sobel_y = cv2.Sobel(aeh_img, cv2.CV_64F, 0, 1, ksize=3)
sobel_xy = np.sqrt(sobel_x ** 2 + sobel_y ** 2)
brain_mask = sys.brain_mask
log_img = exposure.adjust_log(sobel_xy)
otsu = filters.threshold_otsu(log_img[brain_mask != 0])
log_img[log_img < otsu] = 0
def distance(x1, y1, x2, y2):
return np.sqrt((x1 - x2) ** 2 + (y1 - y2) ** 2)
dist_map = np.zeros((h, w))
for row in range(h):
for col in range(w):
if log_img[row, col] == 0:
continue
dist_map[row, col] = distance(x, y, col, row)
maxV = dist_map.max()
minV = dist_map.min()
dist_map[dist_map != 0] = maxV - dist_map[dist_map != 0] # pixel inversion
dist_map = ((dist_map - minV) / (maxV - minV)) * 255
res = dist_map.copy()
maxV = dist_map.max()
minV = dist_map[dist_map != 0].min()
middle = maxV - 1 * (maxV - minV) // 3
print(maxV, minV, middle)
otsu = filters.threshold_otsu(dist_map[dist_map > middle])
dist_map[dist_map < otsu] = 0
# kernel = np.ones((2,2), 'uint8')
# lesion_dilation = cv2.dilate(dist_map, kernel, iterations = 1)
# lesion_erode = cv2.erode(lesion_dilation, kernel, iterations = 1)
# lesion_dilation = cv2.dilate(lesion_erode, kernel, iterations = 1)
ret, mask = cv2.threshold(dist_map, 127, 255, 0)
mask = cv2.drawContours(mask, sys.mask_contours, 0, (255, 0, 0), 1)
# cv2.rectangle(mask, (max(0, x - 2), max(y - 2, 0)),
# (min(256, x + 2),min(256, y + 2)), (227,23,13), 1)
plt.figure(figsize=(15, 7))
plt.subplot(1, 5, 1)
plt.imshow(cbf_img)
plt.title('Input Image')
plt.subplot(1, 5, 2)
plt.imshow(sobel_xy)
plt.title('Sobel Power Image')
plt.subplot(1, 5, 3)
plt.imshow(log_img)
plt.title('Enhance Image with Log')
plt.subplot(1, 5, 4)
plt.imshow(res)
plt.title('Distance Image')
plt.subplot(1, 5, 5)
plt.imshow(mask)
plt.title('Output Image')
plt.show()
return dist_map
plt.imshow(sobel_xx + sobel_yy)
plt.title('xx + yy')
plt.subplot(2, 5, 9)
plt.imshow(lap)
plt.title('Laplacian')
plt.show()
# In[224]:
img = sobel_2xy
plt.figure(figsize=(15, 7))
plt.subplot(1, 2, 1)
plt.imshow(img)
plt.subplot(1, 2, 2)
plt.imshow(exposure.adjust_log(img))
# In[225]:
brain_mask = sys.brain_mask
log_img = exposure.adjust_log(img)
otsu = filters.threshold_otsu(log_img[brain_mask != 0])
otsu
# In[226]:
# img = res.copy()
log_img[img < otsu] = 0
plt.imshow(log_img)
def log_transform(img, gain=1):
return exposure.adjust_log(img, gain)
def adjust_brightness_gamma_log(sink):
sink = exposure.adjust_gamma(sink, 3, 1)
sink = exposure.adjust_log(sink, 2)
return sink
# Display cumulative distribution
img_cdf, bins = exposure.cumulative_distribution(image, bins)
ax_cdf.plot(bins, img_cdf, 'r')
ax_cdf.set_yticks([])
return ax_img, ax_hist, ax_cdf
# Load an example image
img = data.moon()
# Gamma
gamma_corrected = exposure.adjust_gamma(img, 2)
# Logarithmic
logarithmic_corrected = exposure.adjust_log(img, 1)
# Display results
fig = plt.figure(figsize=(8, 5))
axes = np.zeros((2, 3), dtype=np.object)
axes[0, 0] = plt.subplot(2, 3, 1)
axes[0, 1] = plt.subplot(2, 3, 2, sharex=axes[0, 0], sharey=axes[0, 0])
axes[0, 2] = plt.subplot(2, 3, 3, sharex=axes[0, 0], sharey=axes[0, 0])
axes[1, 0] = plt.subplot(2, 3, 4)
axes[1, 1] = plt.subplot(2, 3, 5)
axes[1, 2] = plt.subplot(2, 3, 6)
ax_img, ax_hist, ax_cdf = plot_img_and_hist(img, axes[:, 0])
ax_img.set_title('Low contrast image')
y_min, y_max = ax_hist.get_ylim()