def generate_subRegions(img, patchSize, region_patch_ratio, eraseMap, k, minSize, sigma, thresh=0, radius=220, centers = np.array([219.5, 219.5])):
im = img[:, :, 0]
F0, n_region = segment.segment_label(img, sigma, k, minSize)
eraseLabels = set(list(F0[np.where(eraseMap == 1)].flatten()))
region_patch_list = [[] for i in range(n_region)]
for l in range(n_region):
if l in eraseLabels:
region_patch_list[l] = []
else:
region_patch_centers = list(np.argwhere(F0 == l))
if len(region_patch_centers) == 0:
continue
region_values = im[np.where(F0 == l)]
region_mean = region_values.mean()
# print thresh
if region_mean < thresh:
region_patch_list[l] = []
else:
hw = patchSize / 2
region_patch_gride = np.zeros((len(region_patch_centers), 4))
region_patch_gride[:, :2] = np.array(region_patch_centers) - hw
region_patch_gride[:, 2:] = patchSize
patch_list = list(region_patch_gride)
for ii in range(len(region_patch_centers)):
ll = patch_list[ii]
if np.random.rand(1, )[0] < region_patch_ratio:
if isWithinCircle(ll, centers, radius):
region_patch_list[l].append(ll)
return F0, region_patch_list, eraseLabels
def hierarchical_segmentation(I, k = 100, feature_mask = features.SimilarityMask(1, 1, 1, 1)):
F0, n_region = segment.segment_label(I, 0.8, k, 100)
adj_mat, A0 = _calc_adjacency_matrix(F0, n_region)
feature_extractor = features.Features(I, F0, n_region)
# stores list of regions sorted by their similarity
S = _build_initial_similarity_set(A0, feature_extractor)
# stores region label and its parent (empty if initial).
R = {i : () for i in range(n_region)}
A = [A0] # stores adjacency relation for each step
F = [F0] # stores label image for each step
# greedy hierarchical grouping loop
while len(S):
(s, (i, j)) = S.pop()
t = feature_extractor.merge(i, j)
# record merged region (larger region should come first)
R[t] = (i, j) if feature_extractor.size[j] < feature_extractor.size[i] else (j, i)
Ak = _new_adjacency_dict(A[-1], i, j, t)
A.append(Ak)
S = _merge_similarity_set(feature_extractor, Ak, S, i, j, t)
F.append(_new_label_image(F[-1], i, j, t))
# bounding boxes for each hierarchy
L = feature_extractor.bbox
return (R, F, L)
def detect_regions(imgFile,
eraseMap,
k,
minSize,
sigma,
thresh=0,
imResize=None):
img = skimage.io.imread(imgFile)
if imResize is not None:
img = imresize(img, (imResize, imResize))
if len(img.shape) == 2:
img = skimage.color.gray2rgb(img)
im = img[:, :, 0]
F0, n_region = segment.segment_label(img, sigma, k, minSize)
eraseLabels = set(list(F0[numpy.where(eraseMap == 1)].flatten()))
filterout_labels = []
for l in range(n_region):
region_values = im[np.where(F0 == l)]
region_mean = region_values.mean()
# print thresh
if region_mean < thresh:
filterout_labels.append(l)
detect_regions = numpy.ones(F0.shape).astype(numpy.uint8)
removeLabels = list(eraseLabels) + filterout_labels
for l in removeLabels:
detect_regions[np.where(F0 == l)] = 0
return detect_regions, F0, removeLabels
def hierarchical_segmentation(I,
k=100,
feature_mask=features.SimilarityMask(1, 1, 1,
1)):
F0, n_region = segment.segment_label(I, 0.8, k, 100)
adj_mat, A0 = _calc_adjacency_matrix(F0, n_region)
feature_extractor = features.Features(I, F0, n_region)
# stores list of regions sorted by their similarity
S = _build_initial_similarity_set(A0, feature_extractor)
# stores region label and its parent (empty if initial).
R = {i: () for i in range(n_region)}
A = [A0] # stores adjacency relation for each step
F = [F0] # stores label image for each step
# greedy hierarchical grouping loop
while len(S):
(s, (i, j)) = S.pop()
t = feature_extractor.merge(i, j)
# record merged region (larger region should come first)
R[t] = (
i,
j) if feature_extractor.size[j] < feature_extractor.size[i] else (
j, i)
Ak = _new_adjacency_dict(A[-1], i, j, t)
A.append(Ak)
S = _merge_similarity_set(feature_extractor, Ak, S, i, j, t)
F.append(_new_label_image(F[-1], i, j, t))
# bounding boxes for each hierarchy
L = feature_extractor.bbox
return (R, F, L)
def hierarchical_segmentation(I, k = 100, feature_mask = features.SimilarityMask(1, 1, 1, 1), eraseMap=None):
# F0, n_region = segment.segment_label(I, 0.8, k, 100)
F0, n_region = segment.segment_label(I, 0.5, k, 500)
n_region = F0.max() + 1
# ++ calculate outside region labels
if eraseMap is not None:
eraseLabels = set(list(F0[numpy.where(eraseMap == 1)].flatten()))
else:
eraseLabels = []
adj_mat, A0 = _calc_adjacency_matrix(F0, n_region)
feature_extractor = features.Features(I, F0, n_region)
# stores list of regions sorted by their similarity
S = _build_initial_similarity_set(A0, feature_extractor)
# ++ pop out regions outside circle
if eraseMap is not None:
ii = len(S) - 1
while ii >= 0:
if (S[ii][1][0] in eraseLabels) or (S[ii][1][1] in eraseLabels):
S.pop(ii)
ii -= 1
# stores region label and its parent (empty if initial).
R = {i : () for i in range(n_region)}
A = [A0] # stores adjacency relation for each step
F = [F0] # stores label image for each step
# greedy hierarchical grouping loop
while len(S):
(s, (i, j)) = S.pop()
t = feature_extractor.merge(i, j)
# record merged region (larger region should come first)
R[t] = (i, j) if feature_extractor.size[j] < feature_extractor.size[i] else (j, i)
Ak = _new_adjacency_dict(A[-1], i, j, t)
A.append(Ak)
S = _merge_similarity_set(feature_extractor, Ak, S, i, j, t)
# ++ pop out regions outside circle
if eraseMap is not None:
ii = len(S) - 1
while ii >= 0:
if (S[ii][1][0] in eraseLabels) or (S[ii][1][1] in eraseLabels):
S.pop(ii)
ii -= 1
F.append(_new_label_image(F[-1], i, j, t))
# bounding boxes for each hierarchy
L = feature_extractor.bbox
L_regions = {}
for r, l in R.iteritems():
if r < n_region:
L_regions[r] = [r]
else:
ll = []
if l[0] >= n_region:
ll = ll + L_regions[l[0]]
else:
ll.append(l[0])
if l[1] >= n_region:
ll = ll + L_regions[l[1]]
else:
ll.append(l[1])
L_regions[r] = ll
return (R, F, L, L_regions, eraseLabels)
def generate_subRegions(imgFileOrImg,
patchSize,
region_patch_ratio,
eraseMap,
k,
minSize,
sigma,
radius=220,
centers=np.array([219.5, 219.5]),
thresh=None,
isSaveDetection=False,
diffResolution=False,
returnFilteroutLabels=False):
if isinstance(imgFileOrImg, str):
img = skimage.io.imread(imgFileOrImg)
if len(img.shape) == 2:
img = skimage.color.gray2rgb(img)
else:
img = imgFileOrImg
# im = rgb2gray(img)
# print im.max(), im.min()
im = img[:, :, 0]
if thresh is None:
thresh = 0
F0, n_region = segment.segment_label(img, sigma, k, minSize)
eraseLabels = set(list(F0[numpy.where(eraseMap == 1)].flatten()))
# plt.imshow(eraseMap)
# plt.show()
# print eraseLabels
region_patch_list = [[] for i in range(n_region)]
filterout_labels = []
for l in range(n_region):
if l in eraseLabels:
region_patch_list[l] = []
else:
region_patch_centers = list(np.argwhere(F0 == l))
if len(region_patch_centers) == 0:
continue
region_values = im[np.where(F0 == l)]
region_mean = region_values.mean()
# print thresh
if region_mean < thresh:
region_patch_list[l] = []
filterout_labels.append(l)
else:
hw = patchSize / 2
region_patch_gride = np.zeros((len(region_patch_centers), 4))
if not diffResolution:
region_patch_gride[:, :2] = np.array(
region_patch_centers) - hw
region_patch_gride[:, 2:] = patchSize
patch_list = list(region_patch_gride)
for ii in range(len(region_patch_centers)):
if not diffResolution:
ll = patch_list[ii]
if np.random.rand(1, )[0] < region_patch_ratio:
if diffResolution:
patchSize = np.array(
esg.centerArr2sizeList(
region_patch_centers[ii]))
ll = np.zeros((1, 4))
ll[:, :2] = region_patch_centers[ii]
ll[:, 2:] = patchSize
ll = list(ll)[0]
if esg.isWithinCircle(ll, centers, radius):
region_patch_list[l].append(ll)
if isSaveDetection and (thresh != 0):
folder = '../../Data/Results/regionDetection/'
alpha = 0.6
colors = numpy.random.randint(0, 255, (n_region, 3))
print eraseLabels
for e in eraseLabels:
colors[e] = 0
color_regions_before = colors[F0]
result_before = (color_regions_before * alpha + img *
(1. - alpha)).astype(numpy.uint8)
print filterout_labels
for e in filterout_labels:
colors[e] = 0
color_regions_after = colors[F0]
result_after = (color_regions_after * alpha + img *
(1. - alpha)).astype(numpy.uint8)
imsave(folder + 'before.jpg', result_before)
imsave(folder + 'after.jpg', result_after)
if returnFilteroutLabels:
return F0, region_patch_list, eraseLabels, filterout_labels
else:
return F0, region_patch_list, eraseLabels