import matplotlib.pyplot as plt
from pyxvis.processing.segmentation import seg_bimodal
from pyxvis.features.extraction import extract_features
from pyxvis.io.plots import plot_ellipses_image
img = plt.imread('../images/N0006_0003b.png') # input image with a fruit
R, _, = seg_bimodal(img) # segmentation
fxell = extract_features('ellipse', bw=R) # extraction of elliptical features
print('Elliptical Features:') # show results
print(fxell) # print elliptical features
plot_ellipses_image(img, fxell) # draw ellipse onto image
import matplotlib.pyplot as plt
from skimage.measure import label
from pyxvis.features.extraction import extract_features
fig = plt.figure()
ax = fig.add_subplot(111)
img = plt.imread('../images/N0001_0004b.png')
implot = plt.imshow(img,cmap='gray')
R = img>0.27 # segmentation
L = label(R) # labeling
n = np.max(L) # number of segmented regions
t = 0
T = np.zeros((n,6))
for i in range(n):
R = (L == i)*1 # binary image of object i
f = extract_features('basicgeo',bw=R)
area = f[4]
# recognition of fruits according to the size
if area>14000 and area<21000:
# extract int features only in the segmented region
h = extract_features('basicint',img=img,bw=R)
T[t,:] = h
t = t+1
ax.text(f[1]-20, f[0]+10, str(t), fontsize=12,color='Red')
plt.show()
F = T[0:t,:]
print('Basic Int-Features:')
print(F)
np.save('IntFeatures.npy',F) # save features
fig = plt.figure()
ax = fig.add_subplot(111)
img = plt.imread('../images/N0001_0004b.png')
img[100:399, 750:849] = 0.5
img[500:699, 850:916] = 0.75
img[20:119, 100:399] = 0.6
img[90:156, 1000:1199] = 0.75
implot = plt.imshow(img, cmap='gray')
R = img > 0.27 # segmentation
L = label(R) # labeling
n = np.max(L) # number of segmented objects
t = 0
T = np.zeros((n, 7))
for i in range(n):
R = (L == i) * 1 # binary image of object i
fx = ['basicgeo', 'hugeo']
f = extract_features(fx, bw=R) # feature extraction
area = f[4]
# recognition of fruits according to the size
if area > 10000 and area < 31000:
h = f[18:] # hu moments
T[t, :] = h
t = t + 1
x = round(1000 * h[0]) # first hu moment
ax.text(f[1] - 20, f[0] + 10, str(int(x)), fontsize=12, color='Red')
plt.show()
F = T[0:t, :]
print('Hu Features:')
print(F)
np.save('HuFeatures.npy', F) # save features
from pyxvis.processing.images import gradlog
from skimage.measure import label
from pyxvis.features.extraction import extract_features
img = plt.imread('../images/small_wheel.png') # input image with a defect
(N, M) = img.shape
e = gradlog(img, 1.25, 4 / 250)
L = label(~e) # labeling of objects
n = np.max(L) # number of detected objects
K1 = np.zeros((N, M), dtype=bool)
K2 = np.zeros((N, M), dtype=bool)
# Analysis of each segmented object
for i in range(n):
R = (L == i) # binary image of object i
f = extract_features('basicgeo',
bw=R * 1) # feature extraction for object i
area = f[4]
# recognition of potential defects according to the size
if area > 20 and area < 40:
K1 = np.bitwise_or(K1, R)
i0 = int(round(f[0]))
j0 = int(round(f[1]))
h = int(round(f[2] / 2))
w = int(round(f[3] / 2))
i1 = max(i0 - h, 0)
j1 = max(j0 - w, 0)
i2 = min(i0 + h, N - 1)
j2 = min(j0 + w, M - 1)
I = img[i1:i2, j1:j2]
bw = R[i1:i2, j1:j2]
x = extract_features('contrast', img=I, bw=bw)
fig = plt.figure()
ax = fig.add_subplot(111)
img = plt.imread('../images/N0001_0004b.png')
implot = plt.imshow(img, cmap='gray')
# Segmentation
R = img > 0.27 # thresholding of light objects
L = label(R) # labeling of objects
n = np.max(L) # number of detected objects
T = np.zeros((n, 18)) # features of each object will stored in a row
# Analysis of each segmented object
t = 0 # count of recognized fruits
for i in range(n):
R = (L == i) * 1 # binary image of object i
f = extract_features('basicgeo', bw=R) # feature extraction for object i
area = f[4]
# recognition of fruits according to the size
if area > 14000 and area < 21000:
T[t, :] = f # storing the features of the fruit t
t = t + 1
# labeling each recognized fruit in the plot
ax.text(f[1] - 20, f[0] + 10, str(t), fontsize=12, color='Red')
# Display and save results
plt.show()
F = T[0:t, :]
print('Basic Geo-Features:')
print(F)
np.save('GeoFeatures.npy', F) # save features
from skimage.measure import label
img = plt.imread('../images/small_wheel.png') # input image with a defect
(N, M) = img.shape
e = gradlog(img, 1.25, 4 / 250)
L = label(~e) # labeling of objects
n = np.max(L) # number of detected objects
K1 = np.zeros((N, M), dtype=bool)
K2 = np.zeros((N, M), dtype=bool)
# Analysis of each segmented object
for i in range(n):
R = L == i # binary image of object i
f = extract_features('basicgeo',
bw=R * 1) # feature extraction for object i
area = f[4]
# recognition of potential defects according to the size
if area > 10 and area < 40:
K1 = np.bitwise_or(K1, R)
i0 = int(round(f[0]))
j0 = int(round(f[1]))
h = int(round(f[2] / 2))
w = int(round(f[3] / 2))
i1 = max(i0 - h, 0)
j1 = max(j0 - w, 0)
i2 = min(i0 + h, N - 1)
j2 = min(j0 + w, M - 1)
I = img[i1:i2, j1:j2]
x = extract_features('clp', img=I)