def extr_feat(img):
#imfilt = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
imfilt = img
imLap = cv2.Laplacian(img, cv2.CV_64F)
#hist_vect,bins = np.histogram(imLap.flatten(),bins=50,range=(0,200))
#entropy
entr = np.sum(entropy(imfilt, disk(5))[:])
#SIFTing
#sift = cv2.xfeatures2d.SIFT_create()
#kp, des = sift.detectAndCompute(imfilt,None)
#corners
corns = cv2.cornerHarris(imfilt, 2, 3, 0.04)
numcorns = np.sum((corns > 0.1 * corns.max()).astype(int))
#more edge pixels on left or right
im_siz = imLap.shape
left_weight = np.sum(imLap[:, 0:im_siz[1] / 2])
right_weight = np.sum(imLap[:, im_siz[1] / 2:])
which_side = left_weight / (left_weight + right_weight)
f_vect = np.hstack((entr, numcorns, which_side))
return f_vect
def extr_feat(img):
imfilt = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
imLap = cv2.Laplacian(imfilt, cv2.CV_64F)
#hist_vect,bins = np.histogram(imLap.flatten(),bins=50,range=(0,200))
#for rgb find the mode
med_col = np.array([0, 0, 0])
for cc in range(3):
med_col[cc] = np.array(np.mean(img[:, :, cc]))
mostcol = np.argmax(med_col)
#crop the image
#entropy
entr = np.sum(entropy(imfilt, disk(5))[:])
#SIFTing
#sift = cv2.xfeatures2d.SIFT_create()
#kp, des = sift.detectAndCompute(imfilt,None)
#corners
corns = cv2.cornerHarris(imfilt, 2, 3, 0.04)
numcorns = np.sum((corns > 0.1 * corns.max()).astype(int))
#more edge pixels on left or right
im_siz = imLap.shape
left_weight = np.sum(imLap[:, 0:im_siz[1] / 2])
right_weight = np.sum(imLap[:, im_siz[1] / 2:])
which_side = left_weight / (left_weight + right_weight)
f_vect = np.hstack((entr, med_col[0] / med_col[1], numcorns, which_side))
return f_vect
def feature_1(img):
"""Computes the average local entropy over each pixel of the image.
The local entropy is computed over a disc of radius 5, and is an average of the local entropy over
the RGB channels or greyscale image if 2D
input: a color or greyscale image img.
returns: a list representing the avg local entropy over each of the 3 color channels
"""
import numpy as np
from skimage.filter.rank import entropy
from skimage.morphology import disk
if img.ndim==3:
ent_list = []
for i in [0,1,2]:
ent_array = entropy(img[:,:,i],disk(5))
ent_list.append(np.mean(ent_array))
return ent_list
else:
ent_array = entropy(img,disk(5))
return [np.mean(ent_array)]*3
def calculateEntropy(self):
self.entropy = 0
for i in range(len(self.weights)):
norm_weights = (self.weights[i]-self.weights[i].min())/\
(self.weights[i].max()-self.weights[i].min())
self.entropy += entropy(norm_weights, disk(12)).mean()
# self.entropy += self.entropyF(self.weights[i].sum(axis=0))+self.entropyF(self.weights[i].sum(axis=-1))
self.entropy /= len(self.weights)
return self.entropy
def image_features_entropy(img, maxPixel, num_features,imageSize):
# X is the feature vector with one row of features per image
# consisting of the pixel values a, num_featuresnd our metric
X=np.zeros(num_features, dtype=float)
image = resize(img, (maxPixel, maxPixel))
imag=entropy(image, disk(5))
# Store the rescaled image pixels
X[0:imageSize] = np.reshape(imag,(1, imageSize))
return X
def image_features_entropy(img, maxPixel, num_features,imageSize):
# X is the feature vector with one row of features per image
# consisting of the pixel values a, num_featuresnd our metric
X=np.zeros(num_features, dtype=float)
image = resize(img, (maxPixel, maxPixel))
imag=entropy(image, disk(5))
# Store the rescaled image pixels
X[0:imageSize] = np.reshape(imag,(1, imageSize))
return X
def entropy_image(image, disk_size): """ This function takes in a single-channel image and returns it's entropy within a disk size around it. Inputs: - image: a single-channel image matrix. - disk_size: a pixel radius to compute the entropy around. Outputs: - entropied: a matrix representation of an image To screen: a rainbow-colored image that represents the entropy. """ entropied = entropy(image, disk(disk_size)) imshow(entropied, cmap = plt.cm.jet) return entropied
def test_entropy():
# verify that entropy is coherent with bitdepth of the input data
selem = np.ones((16, 16), dtype=np.uint8)
# 1 bit per pixel
data = np.tile(np.asarray([0, 1]), (100, 100)).astype(np.uint8)
assert(np.max(rank.entropy(data, selem)) == 1)
# 2 bit per pixel
data = np.tile(np.asarray([[0, 1], [2, 3]]), (10, 10)).astype(np.uint8)
assert(np.max(rank.entropy(data, selem)) == 2)
# 3 bit per pixel
data = np.tile(
np.asarray([[0, 1, 2, 3], [4, 5, 6, 7]]), (10, 10)).astype(np.uint8)
assert(np.max(rank.entropy(data, selem)) == 3)
# 4 bit per pixel
data = np.tile(
np.reshape(np.arange(16), (4, 4)), (10, 10)).astype(np.uint8)
assert(np.max(rank.entropy(data, selem)) == 4)
# 6 bit per pixel
data = np.tile(
np.reshape(np.arange(64), (8, 8)), (10, 10)).astype(np.uint8)
assert(np.max(rank.entropy(data, selem)) == 6)
# 8-bit per pixel
data = np.tile(
np.reshape(np.arange(256), (16, 16)), (10, 10)).astype(np.uint8)
assert(np.max(rank.entropy(data, selem)) == 8)
# 12 bit per pixel
selem = np.ones((64, 64), dtype=np.uint8)
data = np.tile(
np.reshape(np.arange(4096), (64, 64)), (2, 2)).astype(np.uint16)
assert(np.max(rank.entropy(data, selem)) == 12)
# make sure output is of dtype double
out = rank.entropy(data, np.ones((16, 16), dtype=np.uint8))
assert out.dtype == np.double
def test_entropy():
# verify that entropy is coherent with bitdepth of the input data
selem = np.ones((16, 16), dtype=np.uint8)
# 1 bit per pixel
data = np.tile(np.asarray([0, 1]), (100, 100)).astype(np.uint8)
assert (np.max(rank.entropy(data, selem)) == 1)
# 2 bit per pixel
data = np.tile(np.asarray([[0, 1], [2, 3]]), (10, 10)).astype(np.uint8)
assert (np.max(rank.entropy(data, selem)) == 2)
# 3 bit per pixel
data = np.tile(np.asarray([[0, 1, 2, 3], [4, 5, 6, 7]]),
(10, 10)).astype(np.uint8)
assert (np.max(rank.entropy(data, selem)) == 3)
# 4 bit per pixel
data = np.tile(np.reshape(np.arange(16), (4, 4)),
(10, 10)).astype(np.uint8)
assert (np.max(rank.entropy(data, selem)) == 4)
# 6 bit per pixel
data = np.tile(np.reshape(np.arange(64), (8, 8)),
(10, 10)).astype(np.uint8)
assert (np.max(rank.entropy(data, selem)) == 6)
# 8-bit per pixel
data = np.tile(np.reshape(np.arange(256), (16, 16)),
(10, 10)).astype(np.uint8)
assert (np.max(rank.entropy(data, selem)) == 8)
# 12 bit per pixel
selem = np.ones((64, 64), dtype=np.uint8)
data = np.tile(np.reshape(np.arange(4096), (64, 64)),
(2, 2)).astype(np.uint16)
assert (np.max(rank.entropy(data, selem)) == 12)
# make sure output is of dtype double
out = rank.entropy(data, np.ones((16, 16), dtype=np.uint8))
assert out.dtype == np.double
def plotEntropy(filename):
"""
Plots the entropy in the data. No really suitable as the input data
not in a useful format.
"""
data = pf.getdata(filename).astype(np.uint32)
#plot the image
fig = plt.figure(figsize=(18, 10))
ax1 = fig.add_subplot(121)
ax2 = fig.add_subplot(122)
ax1.set_title('Data, Single Qaudrant')
ax2.set_title('Background Mask')
im1 = ax1.imshow(np.log10(data), origin='lower', vmin=2., vmax=3.5, interpolation='none')
im2 = ax2.imshow(entropy(data, disk(5)), origin='lower', interpolation='none')
c1 = plt.colorbar(im1, ax=ax1, orientation='horizontal')
c2 = plt.colorbar(im2, ax=ax2, orientation='horizontal')
c1.set_label('$\log_{10}$(Counts [ADU])')
c2.set_label('Entropy')
plt.savefig('entropy.png')
plt.close()
def intensity_stats(im):
if im.shape != (64, 64):
im = transform.resize(im, (64, 64))
l = morphology.disk(10)
h = entropy((im*255.0).astype(np.uint8), l)
return [np.mean(h), np.std(h), float(np.std(im))]
image = img_as_ubyte(original_image)
#preprocess image
#arr = mean(arr, disk(20))
#arr = mean_bilateral(arr.astype(np.uint16), disk(8), s0=4 , s1=4)
#(arr,labels_image,number_regions)=pms.segment(arr,spatial_radius=6,range_radius=4.5,min_density=50)
#plot original image
fig, [(ax0, ax1, ax2),(ax3, ax4, ax5)] = plt.subplots(ncols=3, nrows=2, figsize=(15,8))
img0 = ax0.imshow(image, cmap=plt.cm.gray)
ax0.set_title('Imagen original')
ax0.axis('off')
fig.colorbar(img0, ax=ax0)
#plot entropy image
entro=entropy(arr, disk(3))
img1 = ax1.imshow(entro,cmap=plt.cm.jet)
ax1.set_title('Entropia')
ax1.axis('off')
fig.colorbar(img1, ax=ax1)
#plot histogram of entropy image
ax2.set_title('Histograma')
ax2.hist(entro.ravel(), 256, histtype='step', color='black')
ax2.ticklabel_format(axis='y', style='scientific', scilimits=(0,0))
ax2.set_xlim(0,7)
ax2.set_yticks([])
#plot sample extracted from image
sample_location = (320,240)
Entropy
=======
Image entropy is a quantity which is used to describe the amount of information
coded in an image.
"""
import matplotlib.pyplot as plt
from skimage import data
from skimage.filter.rank import entropy
from skimage.morphology import disk
from skimage.util import img_as_ubyte
image = img_as_ubyte(data.camera())
fig, (ax0, ax1) = plt.subplots(ncols=2, figsize=(10, 4))
img0 = ax0.imshow(image, cmap=plt.cm.gray)
ax0.set_title('Image')
ax0.axis('off')
plt.colorbar(img0, ax=ax0)
img1 = ax1.imshow(entropy(image, disk(5)), cmap=plt.cm.jet)
ax1.set_title('Entropy')
ax1.axis('off')
plt.colorbar(img1, ax=ax1)
plt.show()
def get_entropy_and_mean_motion_temp(videofile):
"""
After messing around with detecting the full body coordinates, I moved to simply calculating aggregate measures of the amount of motion by individuals during a date. This is better because it accounts for factors like head movement or gestures.
It still didn't help much though because how much a person moves during a date doesn't seem to provide much information about how the date will turn out.
"""
# input video to use (called cam b/c I'm lazy, actually a cap "capture")
cam = cv2.VideoCapture(videofile)
# read the first frame to initialize
if cam.isOpened():
print('open')
else:
print('not open')
ret, frame = cam.read()
if not ret:
sys.exit(1)
# get height and width
h, w = reduce_region(frame).shape[:2]
# init the prev frame
prev_frame = reduce_region(frame.copy())
# init motion his to an array of zeros of size (h,w)
motion_history = np.zeros((h, w), np.float32)
sum_history = np.zeros((h, w), np.float32)
hsv = np.zeros((h, w, 3), np.uint8)
hsv[:, :, 1] = 255
# iterate through the frames
frame_num = 0
while True:
# read in a frame
ret, frame = cam.read()
frame_num += 1
if not ret:
break
if frame_num % 5 == 0:
frame = reduce_region(frame)
# calc the difference between this frame and the prev one
frame_diff = cv2.absdiff(frame, prev_frame)
# convert to gray
gray_diff = cv2.cvtColor(frame_diff, cv2.COLOR_BGR2GRAY)
# get the threshold from the user window
thrs = DEFAULT_THRESHOLD
# get the motion_mask value to use based upon the threshold and diff
ret, motion_mask = cv2.threshold(gray_diff, thrs, 1,
cv2.THRESH_BINARY)
sum_history += motion_mask
prev_frame = frame.copy()
cam.release()
cv2.destroyAllWindows()
history_max = sum_history.max(axis=1).max()
entropy_img = entropy(sum_history / history_max, disk(5))
entropy_sum = entropy_img.sum(axis=1).sum()
motion_template_mean = sum_history.mean(axis=1).mean()
print('sum'),
print(entropy_sum)
print('mean'),
print(motion_template_mean)
return entropy_sum, motion_template_mean
thresholding. In practice, you might want to define a region for tinting based on segmentation results or blob detection methods. """ from skimage.filter import rank # Square regions defined as slices over the first two dimensions. top_left = (slice(100),) * 2 bottom_right = (slice(-100, None),) * 2 sliced_image = image.copy() sliced_image[top_left] = colorize(image[top_left], 0.82, saturation=0.5) sliced_image[bottom_right] = colorize(image[bottom_right], 0.5, saturation=0.5) # Create a mask selecting regions with interesting texture. noisy = rank.entropy(grayscale_image, np.ones((9, 9))) textured_regions = noisy > 4 # Note that using `colorize` here is a bit more difficult, since `rgb2hsv` # expects an RGB image (height x width x channel), but fancy-indexing returns # a set of RGB pixels (# pixels x channel). masked_image = image.copy() masked_image[textured_regions, :] *= red_multiplier fig, (ax1, ax2) = plt.subplots(ncols=2, figsize=(8, 4)) ax1.imshow(sliced_image) ax2.imshow(masked_image) plt.show() """ .. image:: PLOT2RST.current_figure
import matplotlib.patches as mp
from skimage.filter.rank import entropy, otsu
from skimage.filter import threshold_otsu
from skimage.morphology import square, rectangle, label, closing, disk, binary_erosion, opening
from skimage.color import label2rgb, rgb2gray
from skimage.segmentation import clear_border
from skimage.measure import regionprops
from skimage import io
i = rgb2gray(io.imread('z.jpg'))
t = threshold_otsu(i)
i = i > t
z = binary_erosion(1 - i, square(3))
z = np.logical_xor(z, i)
z = binary_erosion(1 - z, square(2))
i = 1 - z
i = entropy(i, rectangle(18, 20))
t = threshold_otsu(i)
c = opening(i > t, square(2))
b = c.copy()
clear_border(b)
l = label(b)
z = np.logical_xor(c, b)
l[z] = -1
iol = label2rgb(l, image=i)
fig, ax = plt.subplots(ncols=1, nrows=1, figsize=(6, 6))
ax.imshow(iol)
for region in regionprops(l, ['Area', 'BoundingBox']):
if region['Area'] < 3500:
continue
minr, minc, maxr, maxc = region['BoundingBox']
rect = mp.Rectangle((minc, minr),
def segmentation(sample):
samples = []
cores = []
images = []
cf = .9
entropy_ratio = .5
core_ratio = .07 #around 10% of image is core
eq_thresh = 10
csize = 300 #change it later
# for sample in samples:
# if image_number<5:
# image_number += 1
# continue
# image_number += 1
try:
gray_images = np.array([i for i in open_image(sample)])
except:
print 'can\'t open'
# continue
m, n = gray_images[0].shape
g_min = np.min(gray_images[1:], axis=0)
g_max = np.max(gray_images[1:], axis=0)
g_avg = np.average(gray_images[1:], axis=0)
g_mean = rank.mean_bilateral(g_max, disk(40))
images.append(g_max)
selem = disk(5)
diff = g_max-g_min
diff1 = g_avg-g_min
diff2 = g_max-g_avg
h2 = histogram(diff2)
'''
equalize image -> cores are white or black
'''
equalized = img_as_ubyte(exposure.equalize_hist(diff))
#equalized = img_as_ubyte(exposure.equalize_hist(g_min))#g_min
equalized = exposure.adjust_gamma(g_max,2)
##eq_mask = []
#equalized = img_as_ubyte(exposure.equalize_hist(mask))
#eq_mask = equalized<eq_thresh
'''
local otsu
'''
radius = 20
selem = disk(radius)
local_otsu = rank.otsu(equalized, selem)
# local_otsu = tmp<threshold_otsu(equalized)
bg = diff<=local_otsu
ent = rank.entropy(g_max*~bg, disk(35))
grad = rank.gradient(g_mean, disk(50))
tmp = ent*grad
core_mask = tmp>(np.min(tmp)+(np.max(tmp)-np.min(tmp))*entropy_ratio)
#
# h = histogram(local_otsu)
# cdf = 0
# t = g_min.shape[0]*g_min.shape[1]*core_ratio
#
# for i in range(len(h)):
# cdf += h[i]
# if cdf > t:
# maxi = i
# break
#
# core_mask = (local_otsu<maxi)
## imshow(core_mask)
# ##cores = np.logical_and(eq_mask, core_mask)
# ##imshow(eq_mask)
# #
# #
lbl, num_lbl = ndi.label(core_mask)
for i in range(1,num_lbl+1):
'''
lbl==0 is background
'''
c = np.where(np.max(lbl==i, axis=0)==True)[0]
left = c[0]
right = c[-1]
c = np.where(np.max(lbl==i, axis=1)==True)[0]
up = c[0]
down = c[-1]
# '''
# Don't consider edge cores
# '''
# if left<csize/2 or right>n-csize/2:
# continue
# if up<csize/2 or down>m-csize/2:
# continue
#
core = np.zeros((csize, csize))
h = down-up
w = right-left
middle_x = min(max((up+down)/2, csize/2),m-csize/2)
middle_y = min(max((left+right)/2, csize/2), n-csize/2)
# core = (core_mask*gray_images[0])[middle_x-csize/2:middle_x+csize/2, middle_y-csize/2:middle_y+csize/2]
core = gray_images[0][middle_x-csize/2:middle_x+csize/2, middle_y-csize/2:middle_y+csize/2]
core = exposure.adjust_gamma(core,.5)
cores.append(core)
return cores
# print 'image', image_number
#
#if __name__=='__main__':
# os.system('rm %s -R'%(output_dir))
# os.system('mkdir %s'%(output_dir))
# #os.system('mkdir %sres/021'%(input_dir))
# #os.system('mkdir %sres/041'%(input_dir))
# for i in range(len(cores)):
# image_name = '%s/%i.png'%(output_dir, i)
# imsave(image_name, cores[i])
def get_entropy_and_mean_motion_temp(videofile):
"""
After messing around with detecting the full body coordinates, I moved to simply calculating aggregate measures of the amount of motion by individuals during a date. This is better because it accounts for factors like head movement or gestures.
It still didn't help much though because how much a person moves during a date doesn't seem to provide much information about how the date will turn out.
"""
# input video to use (called cam b/c I'm lazy, actually a cap "capture")
cam = cv2.VideoCapture(videofile)
# read the first frame to initialize
if cam.isOpened():
print('open')
else:
print('not open')
ret, frame = cam.read()
if not ret:
sys.exit(1)
# get height and width
h, w = reduce_region(frame).shape[:2]
# init the prev frame
prev_frame = reduce_region(frame.copy())
# init motion his to an array of zeros of size (h,w)
motion_history = np.zeros((h, w), np.float32)
sum_history = np.zeros((h, w), np.float32)
hsv = np.zeros((h, w, 3), np.uint8)
hsv[:,:,1] = 255
# iterate through the frames
frame_num = 0
while True:
# read in a frame
ret, frame = cam.read()
frame_num += 1
if not ret:
break
if frame_num % 5 == 0:
frame = reduce_region(frame)
# calc the difference between this frame and the prev one
frame_diff = cv2.absdiff(frame, prev_frame)
# convert to gray
gray_diff = cv2.cvtColor(frame_diff, cv2.COLOR_BGR2GRAY)
# get the threshold from the user window
thrs = DEFAULT_THRESHOLD
# get the motion_mask value to use based upon the threshold and diff
ret, motion_mask = cv2.threshold(gray_diff, thrs, 1, cv2.THRESH_BINARY)
sum_history += motion_mask
prev_frame = frame.copy()
cam.release()
cv2.destroyAllWindows()
history_max = sum_history.max(axis=1).max()
entropy_img = entropy(sum_history/history_max, disk(5))
entropy_sum = entropy_img.sum(axis=1).sum()
motion_template_mean = sum_history.mean(axis=1).mean()
print('sum'),
print(entropy_sum)
print('mean'),
print(motion_template_mean)
return entropy_sum, motion_template_mean
# sx = ndimage.sobel(seg, axis=0, mode='constant')
# sy = ndimage.sobel(seg, axis=1, mode='constant')
# contorno = n.hypot(sx, sy)
# #imsave('cont.jpg', contorno)
# area = n.sum(seg != 0)
# peri = n.sum(contorno != 0)
# raz = float(peri**2)/area
# qr.append(raz)
# D2.append(area, peri)
# # guardamos a media das razoes de perimetro/area de todos os segmentos
# # da pintura
# qrm = n.sum(qr) / k
# entropia local (media)
entropia_local = entropy(im_media, disk(5))
sh = entropia_local.shape
m_entropia_local = n.sum(entropia_local) / (sh[0]*sh[1])
qel = m_entropia_local
# entropia local (media) com disco maior, 50x50
entropia_local_maior = entropy(im_media, disk(50))
sh2 = entropia_local_maior.shape
m_entropia_local_maior = n.sum(entropia_local_maior) / (sh2[0]*sh2[1])
qel_maior = m_entropia_local_maior
# usando canny/sobel + transformada de hough para detectar linhas de qualquer angulo
#edges = canny(im_binaria, 2, 1, 25)
#edges = canny(im_binaria, 0.3, 0.2)
sx = ndimage.sobel(im_binaria, axis=0, mode='constant')
from skimage.morphology import disk
import numpy as np
import matplotlib.pyplot as plt
image = data.camera()
plt.figure(figsize=(10, 4))
plt.subplot(1, 2, 1)
plt.imshow(image, cmap=plt.cm.gray)
plt.title('Image')
plt.colorbar()
plt.axis('off')
plt.subplot(1, 2, 2)
plt.imshow(entropy(image, disk(5)), cmap=plt.cm.jet)
plt.title('Entropy')
plt.colorbar()
plt.axis('off')
"""
.. image:: PLOT2RST.current_figure
Implementation
==============
The central part of the `skimage.rank` filters is build on a sliding window
that updates the local gray-level histogram. This approach limits the algorithm
complexity to O(n) where n is the number of image pixels. The complexity is
also limited with respect to the structuring element size.
=======
"""
import numpy as np
import matplotlib.pyplot as plt
from skimage import data
from skimage.filter.rank import entropy
from skimage.morphology import disk
# defining a 8- and a 16-bit test images
a8 = data.camera()
a16 = a8.astype(np.uint16) * 4
ent8 = entropy(a8, disk(5)) # pixel value contain 10x the local entropy
ent16 = entropy(a16, disk(5)) # pixel value contain 1000x the local entropy
# display results
plt.figure(figsize=(10, 10))
plt.subplot(2,2,1)
plt.imshow(a8, cmap=plt.cm.gray)
plt.xlabel('8-bit image')
plt.colorbar()
plt.subplot(2,2,2)
plt.imshow(ent8, cmap=plt.cm.jet)
plt.xlabel('entropy*10')
plt.colorbar()
from skimage import data
from skimage.filter.rank import entropy
from skimage.morphology import disk
import numpy as np
import matplotlib.pyplot as plt
image = data.camera()
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 4))
fig.colorbar(ax1.imshow(image, cmap=plt.cm.gray), ax=ax1)
ax1.set_title('Image')
ax1.axis('off')
fig.colorbar(ax2.imshow(entropy(image, disk(5)), cmap=plt.cm.jet), ax=ax2)
ax2.set_title('Entropy')
ax2.axis('off')
"""
.. image:: PLOT2RST.current_figure
Implementation
==============
The central part of the `skimage.rank` filters is build on a sliding window
that updates the local gray-level histogram. This approach limits the algorithm
complexity to O(n) where n is the number of image pixels. The complexity is
also limited with respect to the structuring element size.
# AC, DTU, April 2014
# Code written in preparation of the J-PARC beamtime
# First step to locate blobs is adaptive equalization
# Source: http://scikit-image.org/docs/0.9.x/auto_examples/plot_equalize.html
import matplotlib.pyplot as plt
import pyfits
from skimage import data
from skimage.filter.rank import entropy
from skimage.morphology import disk
from skimage.util import img_as_ubyte
file = pyfits.open('/Users/Alberto/Desktop/Test_image_raw.fits')
image = file[0].data
fig, (ax0, ax1) = plt.subplots(ncols=2, figsize=(10, 4))
img0 = ax0.imshow(image, cmap=plt.cm.gray)
ax0.set_title('Image')
ax0.axis('off')
plt.colorbar(img0, ax=ax0)
img1 = ax1.imshow(entropy(image, disk(3)), cmap=plt.cm.jet)
ax1.set_title('Entropy')
ax1.axis('off')
plt.colorbar(img1, ax=ax1)
plt.show()
def kmFindEntropy(frame):
frameBlurred = cv2.GaussianBlur(frame, (3,3), 0)
gray = cv2.cvtColor(frameBlurred, cv2.COLOR_BGR2GRAY)
entropyImg = entropy(img_as_ubyte(gray), disk(5))
return entropyImg
from skimage import data
from skimage.filter.rank import entropy
from skimage.morphology import disk
import numpy as np
import matplotlib.pyplot as plt
image = data.camera()
fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(10, 4))
fig.colorbar(ax1.imshow(image, cmap=plt.cm.gray), ax=ax1)
ax1.set_title('Image')
ax1.axis('off')
fig.colorbar(ax2.imshow(entropy(image, disk(5)), cmap=plt.cm.jet), ax=ax2)
ax2.set_title('Entropy')
ax2.axis('off')
"""
.. image:: PLOT2RST.current_figure
Implementation
==============
The central part of the `skimage.rank` filters is build on a sliding window
that updates the local gray-level histogram. This approach limits the algorithm
complexity to O(n) where n is the number of image pixels. The complexity is
also limited with respect to the structuring element size.
In the following we compare the performance of different implementations
# AC, DTU, April 2014
# Code written in preparation of the J-PARC beamtime
# First step to locate blobs is adaptive equalization
# Source: http://scikit-image.org/docs/0.9.x/auto_examples/plot_equalize.html
import pyfits
import matplotlib
import matplotlib.pyplot as plt
import numpy as np
import pylab as py
import matplotlib.pyplot as plt
import matplotlib.cm as cm
from skimage import data, img_as_float
from skimage import exposure
from skimage.filter.rank import entropy
file = pyfits.open('/Users/Alberto/Desktop/Test_image_raw.fits')
img = file[0].data
img1 = ax1.imshow(entropy(img, disk(5)), cmap=plt.cm.jet)
# Conceptually, equalizing and rescaling exposure are the same thing
p2, p98 = np.percentile(img, (2, 98))
img_rescale = exposure.rescale_intensity(img, in_range=(p2, p98))
img_adapteq = exposure.equalize_adapthist(img_rescale, clip_limit=0.1)
#img_eq = exposure.equalize_hist(img_adapteq)
#print img_eq
plt.imshow(img1)
plt.show()
http://www.astro.cornell.edu/research/projects/compression/entropy.html
"""
from skimage.filter.rank import entropy
from skimage.morphology import disk
import numpy as np
import skimage
import os
#Declare here where you want the script to write the csv results
outputfile=open("C
#Provide the directory where all the .jpgs are
filepath="C:
shots=os.listdir(filepath)
#Loop to calculate the average entropy for each image in the directory, and write the "Name","Score" to each row of a csv.
#c is just a counter to print % progess. you can def delete it if you want.
c=0
for shot in shots:
c+=1
# deifning 16-bit images
a16 = skimage.io.imread(filepath+shot,flatten=True).astype(np.uint16)*100
ent16 = entropy(a16,disk(5))
ent=np.average(ent16)
outputfile.write(str(ent)+','+shot+'\n')
print((float(c)/len(shots)*100))
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.patches as mp
from skimage.filter.rank import entropy,otsu
from skimage.filter import threshold_otsu
from skimage.morphology import square,rectangle,label,closing,disk,binary_erosion,opening
from skimage.color import label2rgb,rgb2gray
from skimage.segmentation import clear_border
from skimage.measure import regionprops
from skimage import io
i=rgb2gray(io.imread('z.jpg'))
t=threshold_otsu(i)
i=i>t
z=binary_erosion(1-i,square(3))
i=1-z
i=entropy(i,rectangle(10,1))
t=threshold_otsu(i)
c=i>t
c=closing(c,rectangle(4,28))
b=c.copy()
clear_border(b)
l=label(b)
z=np.logical_xor(c,b)
l[z]=-1
iol=label2rgb(l,image=i)
fig,ax=plt.subplots(ncols=1,nrows=1,figsize=(6,6))
ax.imshow(iol)
for region in regionprops(l,['Area','BoundingBox']):
if region['Area']<3500:
continue
minr,minc,maxr,maxc=region['BoundingBox']
to better use the available image bit, the function returns 10x entropy for
8-bit images and 1000x entropy for 16-bit images.
"""
from skimage import data
from skimage.filter.rank import entropy
from skimage.morphology import disk
import numpy as np
import matplotlib.pyplot as plt
# defining a 8- and a 16-bit test images
a8 = data.camera()
a16 = data.camera().astype(np.uint16) * 4
ent8 = entropy(a8, disk(5)) # pixel value contain 10x the local entropy
ent16 = entropy(a16, disk(5)) # pixel value contain 1000x the local entropy
# display results
plt.figure(figsize=(10, 10))
plt.subplot(2, 2, 1)
plt.imshow(a8, cmap=plt.cm.gray)
plt.xlabel('8-bit image')
plt.colorbar()
plt.subplot(2, 2, 2)
plt.imshow(ent8, cmap=plt.cm.jet)
plt.xlabel('entropy*10')
plt.colorbar()
def preprocessing_gauss_center_entr(img, det):
return im_crop(entropy(ndimage.gaussian_filter((img / MaximumPixelIntensity), sigma=0.95)
, disk(6)), 6.0)
