def get_hog_locations_path(f, data, page_path, number_of_blocks):
""" gets the hog locations from the annotated image, or calculates them, and
them to file, if that is not possible
Note: this is only necessary when some hog features are classified as image """
# If there are images in this page, get their location:
# TODO: Uncommenting this can cause problems due to faulty saved hog
# locations. This should be fixed sometime.
# if data.has_key('hog_locations') and \
# data['hog_locations'].has_key(number_of_blocks):
# return data['hog_locations'][number_of_blocks]
print 'calculating hog locations for %s in mode %s' % (f.name, f.mode)
# otherwise, calculate em, and save em
image = misc.imread(page_path)
hog_locations = calculate_hog_locations(image, number_of_blocks)
# If needed, create the dictionary in 'hog_features'
if not (data.has_key('hog_locations')):
data['hog_locations'] = {}
data['hog_locations'][number_of_blocks] = hog_locations
# write the new data to the original file:
f.seek(0)
f.write(str(data))
# Remove any remaining text from the previous file contents
f.truncate()
return hog_locations
def get_hog_locations_path(f, data, page_path, number_of_blocks):
""" gets the hog locations from the annotated image, or calculates them, and
them to file, if that is not possible
Note: this is only necessary when some hog features are classified as image """
# If there are images in this page, get their location:
# TODO: Uncommenting this can cause problems due to faulty saved hog
# locations. This should be fixed sometime.
# if data.has_key('hog_locations') and \
# data['hog_locations'].has_key(number_of_blocks):
# return data['hog_locations'][number_of_blocks]
print 'calculating hog locations for %s in mode %s' % (f.name, f.mode)
# otherwise, calculate em, and save em
image = misc.imread(page_path)
hog_locations = calculate_hog_locations(image, number_of_blocks)
# If needed, create the dictionary in 'hog_features'
if not(data.has_key('hog_locations')):
data['hog_locations'] = {}
data['hog_locations'][number_of_blocks] = hog_locations
# write the new data to the original file:
f.seek(0)
f.write(str(data))
# Remove any remaining text from the previous file contents
f.truncate()
return hog_locations
def get_hog_features_page(f, data, page_path, number_of_blocks):
""" reads a vector of descriptors for an image. f is the file handle of the
annotated data file, data is the data from that file, page_path the path
to the image"""
block_and_cells = (number_of_blocks, (1, 1))
descriptor = None
if data.has_key('hog_features') and \
data['hog_features'].has_key(block_and_cells):
print 'retrieving hog features for %s' % str(f.name)
descriptor = data['hog_features'][block_and_cells]
else:
print 'calculating hog features for %s' % str(f.name)
# Get the image nparray
image = misc.imread(page_path)
descriptor = calculate_hog(image, number_of_blocks)
# If needed, create the dictionary in 'hog_features'
if not(data.has_key('hog_features')) or \
type(data['hog_features']) != dict:
data['hog_features'] = {}
data['hog_features'][block_and_cells] = descriptor
# write the new data to the original file:
f.seek(0)
f.write(str(data))
# Remove any remaining text from the previous file contents
f.truncate()
assert descriptor != None, 'descriptor cant be None'
# Reshape the descriptors to the pystruct desired shape
# We now have the 0'th index horizontal and the 1'th index
# vertical
descriptor.shape = (number_of_blocks[1], number_of_blocks[0], 8)
# Transpose the first two axes, in order to get from x-y
# coordinates to y-x coordinates
descriptor = descriptor.transpose((1, 0, 2))
return descriptor
def get_hog_features_page(f, data, page_path, number_of_blocks):
""" reads a vector of descriptors for an image. f is the file handle of the
annotated data file, data is the data from that file, page_path the path
to the image"""
block_and_cells = (number_of_blocks, (1, 1))
descriptor = None
if data.has_key('hog_features') and \
data['hog_features'].has_key(block_and_cells):
print 'retrieving hog features for %s' % str(f.name)
descriptor = data['hog_features'][block_and_cells]
else:
print 'calculating hog features for %s' % str(f.name)
# Get the image nparray
image = misc.imread(page_path)
descriptor = calculate_hog(image, number_of_blocks)
# If needed, create the dictionary in 'hog_features'
if not(data.has_key('hog_features')) or \
type(data['hog_features']) != dict:
data['hog_features'] = {}
data['hog_features'][block_and_cells] = descriptor
# write the new data to the original file:
f.seek(0)
f.write(str(data))
# Remove any remaining text from the previous file contents
f.truncate()
assert descriptor != None, 'descriptor cant be None'
# Reshape the descriptors to the pystruct desired shape
# We now have the 0'th index horizontal and the 1'th index
# vertical
descriptor.shape = (number_of_blocks[1], number_of_blocks[0], 8)
# Transpose the first two axes, in order to get from x-y
# coordinates to y-x coordinates
descriptor = descriptor.transpose((1, 0, 2))
return descriptor
def process_felzen(path,ADAP):
if True:
if True:
if False:
im = Image.open(path)
im.thumbnail((im.size[0] / 4, im.size[1] / 4), Image.ANTIALIAS)
im_arr = np.fromstring(im.tobytes(), dtype=np.uint8)
print 'im ', im_arr.size, im.size
im_arr = im_arr.reshape((im.size[1], im.size[0], im_arr.size / (im.size[1] * im.size[0])))
img = im_arr
if True:
img = cv2.imread(
path) # 10978_13_001.png-1.jpg')# densities/143_13_001.png')#558_13_002.png')#fotoscale.jpg')143_13_001.png')#
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
kernel2 = np.ones((9, 9), np.uint8)
data = dict()
for i in gray.ravel():
if data.has_key(i):
data[i] += 1
else:
data[i] = 1
# if data.has_key(255):
max = 0
id = None
for k in data.keys():
# print 'k ',k
if data[k] >= max and k < 240 and k > 130: # 50:#240:
id = k
max = data[k]
#print 'key: ', id, ' - ', max
#id -= 30
THRESH=30
id -= THRESH # 15
if ADAP:
thresh = cv2.adaptiveThreshold(gray, 255,
cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY_INV, 55, 20)
else:
ret, thresh = cv2.threshold(gray, id, 255, cv2.ADAPTIVE_THRESH_MEAN_C + cv2.THRESH_BINARY_INV)
#thresh = cv2.adaptiveThreshold(gray, 255,
# cv2.ADAPTIVE_THRESH_MEAN_C,cv2.THRESH_BINARY_INV,55,20)# + cv2.THRESH_BINARY_INV) # +cv2.THRESH_OTSU)#cv2.ADAPTIVE_THRESH_MEAN_C+
#plt.imshow(thresh)
#plt.show()
kernel = np.ones((3, 3), np.uint8)
opening = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, kernel, iterations=5)
for i in range(opening.shape[0]):
for j in range(opening.shape[1]):
if opening[i][j] == 0:
img[i][j] = (0, 0, 0)
fx = segmentation.felzenszwalb(img, scale=100, sigma=0.8, min_size=50)
ss = set()
for e in range(len(fx)):
for ej in range(len(fx[0])):
ss.add(fx[e][ej])
img2 = np.zeros_like(img)
img2[:, :, 0] = opening
img2[:, :, 1] = opening
img2[:, :, 2] = opening
c = 1
dic = dict()
dic[0] = (0, 0, 0)
for r in range(1, 255, 1):
for r2 in range(1, 255, 1):
for r3 in range(1, 255, 1):
# cols.add((r,r2,r3))
dic[c] = (r, r2, r3)
if c == len(ss) + 5:
break
c += 1
if c == len(ss) + 5:
break
if c == len(ss) + 5:
break
dic[0] = (0, 0, 0)
for e in range(len(fx)):
for ej in range(len(fx[0])):
img2[e][ej] = dic[fx[e][ej]]
#plt.imshow(fx)
#plt.show()
fx = np.uint8(img2)
#s=set()
#for x in fx:
# for y in x:
# print 'x ',y
# s.add((y[0],y[1],y[2]))
#print 'y.s ',len(s)
im = Image.fromarray(fx)
#plt.imshow(im)
#plt.show()
return im
###########
#segments_fz = felzenszwalb(img, scale=100, sigma=0.5, min_size=50)
#ax[0, 0].imshow(mark_boundaries(img, segments_fz))
def process_suzuki(path,THRESH,ADAP):
img = cv2.imread(
path) # 10978_13_001.png-1.jpg')# densities/143_13_001.png')#558_13_002.png')#fotoscale.jpg')143_13_001.png')#
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
kernel2 = np.ones((9, 9), np.uint8)
if ADAP:
thresh = cv2.adaptiveThreshold(gray, 255,
cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY_INV, 55,
20) # + cv2.THRESH_BINARY_INV) # +cv2.THRESH_OTSU)#cv2.ADAPTIVE_THRESH_MEAN_C+
else:
data = dict()
for i in gray.ravel():
if data.has_key(i):
data[i] += 1
else:
data[i] = 1
max = 0
id = None
for k in data.keys():
# print 'k ',k
if data[k] >= max and k < 240 and k > 130: # 50:#240:
id = k
max = data[k]
# print 'key: ', id, ' - ', max
id -= THRESH
ret, thresh = cv2.threshold(gray, id, 255,
cv2.ADAPTIVE_THRESH_MEAN_C + cv2.THRESH_BINARY_INV) # +cv2.THRESH_OTSU)#cv2.ADAPTIVE_THRESH_MEAN_C+
kernel = np.ones((3, 3), np.uint8)
opening = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, kernel, iterations=5)
opening2 = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, kernel, iterations=5)
#plt.imshow(opening)
#plt.show()
_, contours, hierarchy = cv2.findContours(opening.copy(), cv2.RETR_TREE, cv2.CHAIN_APPROX_NONE)
img2 = np.zeros_like(img)
img2[:, :, 0] = opening
img2[:, :, 1] = opening
img2[:, :, 2] = opening
cp=10000
cols = list()
cols2 = set()
for r in range(1, 255, 1):
for r1 in range(1, 255, 1):
for r2 in range(1, 255, 1):
cols.append((r, r1, r2))
cp-=1
if cp==0:
break
if cp == 0:
break
if cp == 0:
break
ind=0
# print len(contours)
shuffle(cols)
for c in cols:
cv2.drawContours(img2, contours, ind, c, -1)
ind += 1
if ind==len(contours):
break
for x in range(opening.shape[0]):
for y in range(opening.shape[1]):
if opening[x][y]==0:
img2[x][y]=(0,0,0)
# plt.imshow(img2)
# plt.show()
#markers = np.uint8(opening)
im = Image.fromarray(img2)
# plt.imshow(im)
# plt.show()
return im
def process_water(path,THRESH,ADAP,DIST_PERC):
img = cv2.imread(path)#10978_13_001.png-1.jpg')# densities/143_13_001.png')#558_13_002.png')#fotoscale.jpg')143_13_001.png')#
gray = cv2.cvtColor(img,cv2.COLOR_BGR2GRAY)
kernel2 = np.ones((9, 9), np.uint8)
#plt.matshow(gray,cmap='gray')
#plt.show()
if ADAP:
thresh = cv2.adaptiveThreshold(gray, 255,
cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY_INV, 55, 20)
else:
data = dict()
for i in gray.ravel():
if data.has_key(i):
data[i] += 1
else:
data[i] = 1
max = 0
id = None
for k in data.keys():
# print 'k ',k
if data[k] >= max and k < 240 and k > 130: # 50:#240:
id = k
max = data[k]
print 'key: ', id, ' - ', max
id -= THRESH # 15
ret, thresh = cv2.threshold(gray,id,255,cv2.ADAPTIVE_THRESH_MEAN_C+cv2.THRESH_BINARY_INV)#+cv2.THRESH_OTSU)#cv2.ADAPTIVE_THRESH_MEAN_C+
#plt.matshow(thresh,cmap='gray')
#plt.show()
kernel = np.ones((3,3),np.uint8)
opening = cv2.morphologyEx(thresh,cv2.MORPH_OPEN,kernel, iterations = 5)
#plt.matshow(opening,cmap='gray')
#plt.show()
sure_bg = cv2.dilate(opening,kernel,iterations=3)
dist_transform = cv2.distanceTransform(opening,cv2.DIST_L2,5)#OPENING
#plt.matshow(dist_transform)
#plt.show()
#print dist_transform.min(), dist_transform.max()
ret, sure_fg = cv2.threshold(dist_transform,int(dist_transform.max()*DIST_PERC),255,0)
sure_fg = cv2.erode(sure_fg,kernel,iterations=3)#3
sure_fg = np.uint8(sure_fg)
unknown = cv2.subtract(sure_bg,sure_fg)
ret, markers = cv2.connectedComponents(sure_fg)
#plt.matshow(markers)
#plt.show()
# Add one to all labels so that sure background is not 0, but 1
markers = markers+1
markers[unknown==255] = 0
markers = cv2.watershed(img,markers)
img[markers == -1] = [0,0,0]
#plt.imshow(markers)
#plt.show()
ss = set()
for e in range(len(markers)):
for ej in range(len(markers[0])):
if markers[e][ej]==-1 or markers[e][ej]==1:
markers[e][ej]=0
#if markers[e][ej]==0:
# img2[e][ej]=(0,0,0)
ss.add(markers[e][ej])
#cols=list()
img2 = np.zeros_like(img)
img2[:, :, 0] = opening
img2[:, :, 1] = opening
img2[:, :, 2] = opening
c=1
dic=dict()
dic[0]=(0,0,0)
for r in range(1,255,1):
for r2 in range(1, 255, 1):
for r3 in range(1, 255, 1):
#cols.add((r,r2,r3))
dic[c]=(r,r2,r3)
if c==len(ss)+5:
break
c+=1
if c==len(ss)+5:
break
if c == len(ss)+5:
break
for e in range(len(markers)):
for ej in range(len(markers[0])):
img2[e][ej]=dic[markers[e][ej]]
#plt.imshow(markers)
#plt.show()
markers = np.uint8(img2)
im = Image.fromarray(markers)
#print im
#print 'mar ', len(ss)
return im
def process_quick(path,ADAP):
if True:
im = Image.open(path)
im.thumbnail((im.size[0]/4,im.size[1]/4), Image.ANTIALIAS)
im_arr = np.fromstring(im.tobytes(), dtype=np.uint8)
print 'im ',im_arr.size, im.size
im_arr = im_arr.reshape((im.size[1], im.size[0],im_arr.size/(im.size[1]*im.size[0])))
img=im_arr
# plt.imshow(img)
# plt.show()
#img = cv2.imread(
# path) # 10978_13_001.png-1.jpg')# densities/143_13_001.png')#558_13_002.png')#fotoscale.jpg')143_13_001.png')#
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY)
kernel2 = np.ones((9, 9), np.uint8)
data = dict()
for i in gray.ravel():
if data.has_key(i):
data[i] += 1
else:
data[i] = 1
max = 0
id = None
for k in data.keys():
# print 'k ',k
if data[k] >= max and k < 240 and k>130: # 50:#240:
id = k
max = data[k]
#print 'key: ', id, ' - ', max
id -= 30
if ADAP:
thresh = cv2.adaptiveThreshold(gray, 255,
cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY_INV, 55, 20)
else:
ret, thresh = cv2.threshold(gray, id, 255,
cv2.ADAPTIVE_THRESH_MEAN_C + cv2.THRESH_BINARY_INV) # +cv2.THRESH_OTSU)#cv2.ADAPTIVE_THRESH_MEAN_C+
kernel = np.ones((3, 3), np.uint8)
opening = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, kernel, iterations=1)
for i in range(opening.shape[0]):
for j in range(opening.shape[1]):
if opening[i][j] == 0:
img[i][j] = (0, 0, 0)
#plt.imshow(img)
#plt.show()
fx = segmentation.quickshift(img,kernel_size=3) # , markers)
#plt.imshow(fx)
#plt.show()
ss = set()
for i in range(fx.shape[0]):
for j in range(fx.shape[1]):
if isinstance(fx[i][j],int):
ss.add(fx[i][j])
else:
ss.add((fx[i][j][0],fx[i][j][1] ,fx[i][j][2]))
#print 'ssle ',len(ss)
c=0
cols=list()
for r in range(1,250,1):
for r2 in range(1, 250, 1):
for r3 in range(1, 250, 1):
cols.append((r,r2,r3))
print 'cols',len(cols)
shuffle(cols)
img2 = np.zeros_like(img)
img2[:, :, 0] = opening
img2[:, :, 1] = opening
img2[:, :, 2] = opening
fx = color.label2rgb(fx, img2, colors=cols, kind='overlay')
for i in range(fx.shape[0]):
for j in range(fx.shape[1]):
if opening[i][j]==0:
fx[i][j]=(0,0,0)
# ss.add((fx[i][j][0], fx[i][j][1], fx[i][j][2]))
#plt.imshow(fx)
#plt.show()
#ss = set()
#for i in range(fx.shape[0]):
# for j in range(fx.shape[1]):
# ss.add((fx[i][j][0], fx[i][j][1], fx[i][j][2]))
#print 'ssle22 ', len(ss)
#plt.imshow(fx)
#plt.show()
fx = np.uint8(fx)
im = Image.fromarray(fx)
#plt.imshow(im)
#plt.show()
return im
#kernel = np.ones((3, 3), np.uint8)
#kernel2 = np.ones((9, 9), np.uint8)
#erosion = cv2.erode(im_arr, kernel, iterations=1)
#dilate = cv2.dilate(erosion, kernel2, iterations=1)
# plt.imshow(dilate)
# plt.show()
#imgM = dilate#Image.fromarray(dilate)
#grayM = cv2.cvtColor(imgM,cv2.COLOR_BGR2GRAY)#cv2.imread('/home/olusiak/water_coins.jpg',0)
#equ = cv2.equalizeHist(gray)
#gray=equ
kernel2 = np.ones((9, 9), np.uint8)
data=dict()
for i in gray.ravel():
if data.has_key(i):
data[i]+=1
else:
data[i]=1
#if data.has_key(255):
max=0
id=None
for k in data.keys():
#print 'k ',k
if data[k]>=max and k<240:#50:#240:
id=k
max=data[k]
print 'key: ',id,' - ',max
id-=15
#plt.imshow(img)
def process(path, N_SEGM, THRESH, ADAP):
im = Image.open(path) #'/home/olusiak/Obrazy/rois/41136_001.png-2.jpg')
#ran=1
#im_arr2 = np.fromstring(im.tobytes(), dtype=np.uint8)
#im_arr2 = im_arr2.reshape((im.size[1], im.size[0], 3))
#for x in range(len(im_arr2)):
# for y in range(len(im_arr2[0])):
# im_arr2[x][y][2]=0#=(im_arr2[0],255,255)
# im_arr2[x][y][1]=0
#plt.imshow(im_arr2)
#plt.show()
im_arr = np.fromstring(im.tobytes(), dtype=np.uint8)
im_arr = im_arr.reshape((im.size[1], im.size[0], 3))
gray = cv2.cvtColor(im_arr, cv2.COLOR_BGR2GRAY)
if ADAP:
thresh = cv2.adaptiveThreshold(gray, 255, cv2.ADAPTIVE_THRESH_MEAN_C,
cv2.THRESH_BINARY_INV, 55, 20)
else:
data = dict()
for i in gray.ravel():
if data.has_key(i):
data[i] += 1
else:
data[i] = 1
# if data.has_key(255):
max = 0
id = None
for k in data.keys():
# print 'k ',k
if data[k] >= max and k < 240 and k > 130: # 50:#240:
id = k
max = data[k]
# for k in data.keys():
# print 'k ',k,' ',data[k]
# print 'key: ',id,' - ',max
id -= THRESH # 35
ret, thresh = cv2.threshold(
gray, id, 255, cv2.ADAPTIVE_THRESH_MEAN_C + cv2.THRESH_BINARY_INV
) #+cv2.THRESH_OTSU)#cv2.ADAPTIVE_THRESH_MEAN_C+
#plt.matshow(thresh,cmap='gray')
#plt.show()
kernel = np.ones((3, 3), np.uint8)
opening = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, kernel, iterations=3)
opening2 = cv2.morphologyEx(thresh, cv2.MORPH_OPEN, kernel, iterations=3)
#plt.imshow(opening)
#plt.show()
for i in range(opening.shape[0]):
for j in range(opening.shape[1]):
if opening[i][j] == 0:
im_arr[i][j] = (0, 0, 0)
#plt.imshow(im_arr)
#plt.show()
#plt.matshow(opening)
#plt.show()
#plt.imshow(im_arr)
#plt.show()
#im = Image.fromarray(im_arr)
#im.thumbnail((im.size[0] / ran, im.size[1] / ran), Image.ANTIALIAS)
#im_arr = np.fromstring(im.tobytes(), dtype=np.uint8)
#im_arr = im_arr.reshape((im.size[1], im.size[0], 3))
#im = Image.fromarray(opening)
#im.thumbnail((im.size[0] / ran, im.size[1] / ran), Image.ANTIALIAS)
#opening = np.fromstring(im.tobytes(), dtype=np.uint8)
#opening = opening.reshape((im.size[1], im.size[0]))
img2 = im_arr
edges = filters.sobel(color.rgb2gray(img2))
labels = segmentation.slic(img2, compactness=30,
n_segments=N_SEGM) #30 2000
#g = graph.rag_mean_color(img2, labels)#added
g = graph.rag_boundary(labels, edges) #first
#graph.show_rag(labels, g, img)
#plt.title('Initial RAG')
labels2 = graph.merge_hierarchical(
labels,
g,
thresh=0.98,
rag_copy=False, # 0.08
in_place_merge=True,
merge_func=merge_boundary,
weight_func=weight_boundary)
#final_labels = graph.cut_threshold(labels, g, 29)#added
#final_label_rgb = color.label2rgb(final_labels, img2, colors=cols, kind='overlay')#added
#labels2=final_label_rgb#added
#plt.imshow(final_label_rgb)
#plt.show()
#graph.show_rag(labels, g, im)
#plt.title('RAG after hierarchical merging')
#plt.figure()
#ret, opening = cv2.threshold(opening,0,255,cv2.THRESH_OTSU)#cv2.ADAPTIVE_THRESH_MEAN_C+
#out = color.label2rgb(labels2, img2, kind='avg')
s = set()
for row in labels2:
for e in row:
s.add(e)
#print 'sss ',len(s)
cols = list()
c = 0
cp = len(s) + 5
for r in range(0, 256, 1):
for r2 in range(0, 256, 1):
for r3 in range(0, 256, 1):
cols.append((r, r2, r3))
cp -= 1
if cp == 0:
break
if cp == 0:
break
if cp == 0:
break
# print 'cols', len(cols)
shuffle(cols)
img2 = np.zeros_like(img2)
img2[:, :, 0] = opening2
img2[:, :, 1] = opening2
img2[:, :, 2] = opening2
out = color.label2rgb(labels2, img2, colors=cols, kind='overlay', alpha=1)
for i in range(img2.shape[0]):
for j in range(img2.shape[1]):
if img2[i][j][0] == 0 and img2[i][j][1] == 0 and img2[i][j][
2] == 0: #,0,0]:
out[i][j][0] = 0
out[i][j][1] = 0
out[i][j][2] = 0
#print 'OUT'
#plt.imshow(out)
#plt.show()
#plt.imshow(out)
#plt.show()
#xx = set()
#for i in range(out.shape[0]): # fx.shape[0]):
# for j in range(out.shape[1]): # fx.shape[1]):
# s = (out[i, j][0], out[i, j][1], out[i, j][2])
# xx.add(s)
out = np.uint8(out)
im = Image.fromarray(out)
#plt.imshow(out)
#plt.show()
return im