def find_concetric_circles(gray_img, min_ring_count=3, visual_debug=False):
# get threshold image used to get crisp-clean edges using blur to remove small features
edges = cv2.adaptiveThreshold(cv2.blur(gray_img, (3, 3)), 255,
cv2.ADAPTIVE_THRESH_MEAN_C,
cv2.THRESH_BINARY, 5, 11)
_, contours, hierarchy = cv2.findContours(edges,
mode=cv2.RETR_TREE,
method=cv2.CHAIN_APPROX_NONE,
offset=(0, 0)) #TC89_KCOS
if visual_debug is not False:
cv2.drawContours(visual_debug, contours, -1, (200, 0, 0))
if contours is None or hierarchy is None:
return []
clusters = get_nested_clusters(contours,
hierarchy[0],
min_nested_count=min_ring_count)
concentric_cirlce_clusters = []
#speed up code by caching computed ellipses
ellipses = {}
# for each cluster fit ellipses and cull members that dont have good ellipse fit
for cluster in clusters:
if visual_debug is not False:
cv2.drawContours(visual_debug, [contours[i] for i in cluster], -1,
(0, 0, 255))
candidate_ellipses = []
for i in cluster:
c = contours[i]
if len(c) > 5:
if not i in ellipses:
e = cv2.fitEllipse(c)
fit = max(dist_pts_ellipse(e, c))
ellipses[i] = e, fit
else:
e, fit = ellipses[i]
a, b = e[1][0] / 2., e[1][1] / 2.
if fit < max(2, max(e[1]) / 20):
candidate_ellipses.append(e)
if visual_debug is not False:
cv2.ellipse(visual_debug, e, (0, 255, 0), 1)
if candidate_ellipses:
cluster_center = np.mean(np.array(
[e[0] for e in candidate_ellipses]),
axis=0)
candidate_ellipses = [
e for e in candidate_ellipses
if np.linalg.norm(e[0] -
cluster_center) < max(3,
min(e[1]) / 20)
]
if len(candidate_ellipses) >= min_ring_count:
concentric_cirlce_clusters.append(candidate_ellipses)
if visual_debug is not False:
cv2.ellipse(visual_debug, candidate_ellipses[-1],
(0, 255, 255), 4)
#return clusters sorted by size of outmost cirlce biggest first.
return sorted(concentric_cirlce_clusters, key=lambda e: -max(e[-1][1]))
def find_concetric_circles(gray_img,min_ring_count=3, visual_debug=False):
# get threshold image used to get crisp-clean edges using blur to remove small features
edges = cv2.adaptiveThreshold(cv2.blur(gray_img,(3,3)), 255, cv2.ADAPTIVE_THRESH_MEAN_C, cv2.THRESH_BINARY, 5, 11)
contours, hierarchy = cv2.findContours(edges,
mode=cv2.RETR_TREE,
method=cv2.CHAIN_APPROX_NONE,offset=(0,0)) #TC89_KCOS
if visual_debug is not False:
cv2.drawContours(visual_debug,contours,-1,(200,0,0))
if contours is None or hierarchy is None:
return []
clusters = get_nested_clusters(contours,hierarchy[0],min_nested_count=min_ring_count)
concentric_cirlce_clusters = []
#speed up code by caching computed ellipses
ellipses = {}
# for each cluster fit ellipses and cull members that dont have good ellipse fit
for cluster in clusters:
if visual_debug is not False:
cv2.drawContours(visual_debug, [contours[i] for i in cluster],-1, (0,0,255))
candidate_ellipses = []
for i in cluster:
c = contours[i]
if len(c)>5:
if not ellipses.has_key(i):
e = cv2.fitEllipse(c)
fit = max(dist_pts_ellipse(e,c))
ellipses[i] = e,fit
else:
e,fit = ellipses[i]
a,b = e[1][0]/2.,e[1][1]/2.
if fit<max(2,max(e[1])/20):
candidate_ellipses.append(e)
if visual_debug is not False:
cv2.ellipse(visual_debug, e, (0,255,0),1)
if candidate_ellipses:
cluster_center = np.mean(np.array([e[0] for e in candidate_ellipses]),axis=0)
candidate_ellipses = [e for e in candidate_ellipses if np.linalg.norm(e[0]-cluster_center)<max(3,min(e[1])/20) ]
if len(candidate_ellipses) >= min_ring_count:
concentric_cirlce_clusters.append(candidate_ellipses)
if visual_debug is not False:
cv2.ellipse(visual_debug, candidate_ellipses[-1], (0,255,255),4)
#return clusters sorted by size of outmost cirlce biggest first.
return sorted(concentric_cirlce_clusters,key=lambda e:-max(e[-1][1]))
def find_concentric_circles(
edge,
scale,
img_contrast,
found_pos,
found_size,
first_check=True,
min_ellipses_num=2,
):
if first_check:
concentric_circle_clusters = []
# OpenCV version compatibility note:
# - Opencv3 returns three results: image, contours, hierarchy
# - Opencv4 returns two results: contours, hierarchy
# We do not use `image` in any case. Therefore, we can ensure
# compatibility by using `*_`
*_, contours, hierarchy = cv2.findContours(
edge, mode=cv2.RETR_TREE, method=cv2.CHAIN_APPROX_TC89_KCOS)
# We use CHAIN_APPROX_TC89_KCOS because it is faster than
# the default method CV_CHAIN_APPROX_NONE
if contours is None or hierarchy is None:
return []
clusters = get_nested_clusters(contours, hierarchy[0],
min_ellipses_num)
# speed up code by caching computed ellipses
ellipses = {}
for cluster in clusters:
candidate_ellipses = []
first_ellipse = True
for i in cluster:
c = contours[i]
if len(c) > 100:
continue
if i in ellipses:
e, fit = ellipses[i]
else:
if len(c) >= 5:
e = cv2.fitEllipse(c)
# Discard duplicates
if first_ellipse:
duplicates = [
k for k in range(len(found_pos))
if LA.norm(e[0] -
found_pos[k]) < found_size[k] +
min(e[1])
]
if len(duplicates) > 0:
ellipses[i] = e, 100.0
break
fit = 0
else:
fit = max(dist_pts_ellipse(e, c)) if min(
e[1]) else 0.0
e = e if min(e[1]) else (e[0], (0.1, 0.1), e[2])
else:
e_center = (
float(c[len(c) // 2][0][0]),
float(c[len(c) // 2][0][1]),
)
e = (e_center, (0.1, 0.1), 0.0)
# Discard duplicates
if first_ellipse:
duplicates = [
k for k in range(len(found_pos))
if LA.norm(e_center -
found_pos[k]) < found_size[k] + 1
]
if len(duplicates) > 0:
ellipses[i] = e, 100.0
break
fit = 0
ellipses[i] = e, fit
# Discard the contour which does not fit the ellipse so well
if first_ellipse or fit < max(1, max(e[1]) / 50):
e = (e[0], e[1], e[2], i)
candidate_ellipses.append(e)
first_ellipse = False
# Discard false positives
if len(candidate_ellipses) < min_ellipses_num:
continue
# Discard the ellipses whose center is far away from the center of the second innermost ellipse
cluster_center = np.array(candidate_ellipses[1][0])
if max(candidate_ellipses[-1][1]) > 200:
candidate_ellipses = [
e for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(e[1]) / 5
]
elif max(candidate_ellipses[-1][1]) > 100:
candidate_ellipses = [
e for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(e[1]) / 10
]
else:
candidate_ellipses = [
e for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(max(e[1]) / 20, 3)
]
# Discard false positives
if len(candidate_ellipses) < min_ellipses_num:
continue
c = contours[candidate_ellipses[-1][3]]
boundary = (
(np.amin(c, axis=0)[0][0], np.amax(c, axis=0)[0][0]),
(np.amin(c, axis=0)[0][1], np.amax(c, axis=0)[0][1]),
)
candidate_ellipses = [(e[0], e[1], e[2])
for e in candidate_ellipses]
concentric_circle_clusters.append((candidate_ellipses, boundary))
# Return clusters sorted by the number of ellipses and the size of largest ellipse
return sorted(concentric_circle_clusters,
key=lambda x: (-len(x[0]), -max(x[0][-1][1])))
else:
*_, contours, hierarchy = cv2.findContours(
edge, mode=cv2.RETR_TREE, method=cv2.CHAIN_APPROX_NONE)
if contours is None or hierarchy is None:
return []
clusters = get_nested_clusters(contours, hierarchy[0],
min_ellipses_num)
# speed up code by caching computed ellipses
ellipses = {}
for cluster in clusters:
candidate_ellipses = []
first_ellipse = True
for i in cluster:
c = contours[i]
if i in ellipses:
e, fit = ellipses[i]
else:
if len(c) >= 5:
e = cv2.fitEllipse(c)
fit = max(dist_pts_ellipse(e, c)) if min(e[1]) else 0.0
if min(e[1]) == 0:
e = (e[0], (e[1][0] + 1.0, e[1][1] + 1.0), e[2])
else:
fit = 0
e_center = (
float(c[len(c) // 2][0][0]),
float(c[len(c) // 2][0][1]),
)
e_size = (
float(abs(c[-1][0][0] - c[0][0][0]) + 1),
float(abs(c[-1][0][1] - c[0][0][1]) + 1),
)
e = (e_center, e_size, 0)
ellipses[i] = e, fit
# Discard the contour which does not fit the ellipse so well
if first_ellipse:
fit_thres = 0.5 + (256 - img_contrast) / 256
else:
if img_contrast <= 96:
fit_thres = max(
e[1]) * scale / 10 + (256 - img_contrast) / 256
else:
fit_thres = max(0.5, max(e[1]) * scale / 10)
if fit < fit_thres:
candidate_ellipses.append(e)
if len(candidate_ellipses) == 4:
break
first_ellipse = False
# Discard false positives
if len(candidate_ellipses) < min_ellipses_num:
continue
# Discard the ellipses whose center is far away from the center of the innermost ellipse
cluster_center = np.array(candidate_ellipses[0][0])
if max(candidate_ellipses[-1][1]) * scale > 200:
candidate_ellipses = [
e for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(e[1]) / 5
]
elif max(candidate_ellipses[-1][1]) * scale > 100:
candidate_ellipses = [
e for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(e[1]) / 10
]
else:
candidate_ellipses = [
e for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(max(e[1]) / 20, 2)
]
# Discard false positives
if len(candidate_ellipses) < min_ellipses_num:
continue
return [(candidate_ellipses, [[0, 0], [0, 0]])]
return []
def find_concentric_circles(
edge,
scale,
img_contrast,
found_pos,
found_size,
first_check=True,
min_ellipses_num=2,
):
if first_check:
concentric_circle_clusters = []
# OpenCV version compatibility note:
# - Opencv3 returns three results: image, contours, hierarchy
# - Opencv4 returns two results: contours, hierarchy
# We do not use `image` in any case. Therefore, we can ensure
# compatibility by using `*_`
*_, contours, hierarchy = cv2.findContours(
edge, mode=cv2.RETR_TREE, method=cv2.CHAIN_APPROX_TC89_KCOS
)
# We use CHAIN_APPROX_TC89_KCOS because it is faster than
# the default method CV_CHAIN_APPROX_NONE
if contours is None or hierarchy is None:
return []
clusters = get_nested_clusters(contours, hierarchy[0], min_ellipses_num)
# speed up code by caching computed ellipses
ellipses = {}
for cluster in clusters:
candidate_ellipses = []
first_ellipse = True
for i in cluster:
c = contours[i]
if len(c) > 100:
continue
if i in ellipses:
e, fit = ellipses[i]
else:
if len(c) >= 5:
e = cv2.fitEllipse(c)
# Discard duplicates
if first_ellipse:
duplicates = [
k
for k in range(len(found_pos))
if LA.norm(e[0] - found_pos[k])
< found_size[k] + min(e[1])
]
if len(duplicates) > 0:
ellipses[i] = e, 100.0
break
fit = 0
else:
fit = max(dist_pts_ellipse(e, c)) if min(e[1]) else 0.0
e = e if min(e[1]) else (e[0], (0.1, 0.1), e[2])
else:
e_center = (
float(c[len(c) // 2][0][0]),
float(c[len(c) // 2][0][1]),
)
e = (e_center, (0.1, 0.1), 0.0)
# Discard duplicates
if first_ellipse:
duplicates = [
k
for k in range(len(found_pos))
if LA.norm(e_center - found_pos[k]) < found_size[k] + 1
]
if len(duplicates) > 0:
ellipses[i] = e, 100.0
break
fit = 0
ellipses[i] = e, fit
# Discard the contour which does not fit the ellipse so well
if first_ellipse or fit < max(1, max(e[1]) / 50):
e = (e[0], e[1], e[2], i)
candidate_ellipses.append(e)
first_ellipse = False
# Discard false positives
if len(candidate_ellipses) < min_ellipses_num:
continue
# Discard the ellipses whose center is far away from the center of the second innermost ellipse
cluster_center = np.array(candidate_ellipses[1][0])
if max(candidate_ellipses[-1][1]) > 200:
candidate_ellipses = [
e
for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(e[1]) / 5
]
elif max(candidate_ellipses[-1][1]) > 100:
candidate_ellipses = [
e
for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(e[1]) / 10
]
else:
candidate_ellipses = [
e
for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(max(e[1]) / 20, 3)
]
# Discard false positives
if len(candidate_ellipses) < min_ellipses_num:
continue
c = contours[candidate_ellipses[-1][3]]
boundary = (
(np.amin(c, axis=0)[0][0], np.amax(c, axis=0)[0][0]),
(np.amin(c, axis=0)[0][1], np.amax(c, axis=0)[0][1]),
)
candidate_ellipses = [(e[0], e[1], e[2]) for e in candidate_ellipses]
concentric_circle_clusters.append((candidate_ellipses, boundary))
# Return clusters sorted by the number of ellipses and the size of largest ellipse
return sorted(
concentric_circle_clusters, key=lambda x: (-len(x[0]), -max(x[0][-1][1]))
)
else:
*_, contours, hierarchy = cv2.findContours(
edge, mode=cv2.RETR_TREE, method=cv2.CHAIN_APPROX_NONE
)
if contours is None or hierarchy is None:
return []
clusters = get_nested_clusters(contours, hierarchy[0], min_ellipses_num)
# speed up code by caching computed ellipses
ellipses = {}
for cluster in clusters:
candidate_ellipses = []
first_ellipse = True
for i in cluster:
c = contours[i]
if i in ellipses:
e, fit = ellipses[i]
else:
if len(c) >= 5:
e = cv2.fitEllipse(c)
fit = max(dist_pts_ellipse(e, c)) if min(e[1]) else 0.0
if min(e[1]) == 0:
e = (e[0], (e[1][0] + 1.0, e[1][1] + 1.0), e[2])
else:
fit = 0
e_center = (
float(c[len(c) // 2][0][0]),
float(c[len(c) // 2][0][1]),
)
e_size = (
float(abs(c[-1][0][0] - c[0][0][0]) + 1),
float(abs(c[-1][0][1] - c[0][0][1]) + 1),
)
e = (e_center, e_size, 0)
ellipses[i] = e, fit
# Discard the contour which does not fit the ellipse so well
if first_ellipse:
fit_thres = 0.5 + (256 - img_contrast) / 256
else:
if img_contrast <= 96:
fit_thres = max(e[1]) * scale / 10 + (256 - img_contrast) / 256
else:
fit_thres = max(0.5, max(e[1]) * scale / 10)
if fit < fit_thres:
candidate_ellipses.append(e)
if len(candidate_ellipses) == 4:
break
first_ellipse = False
# Discard false positives
if len(candidate_ellipses) < min_ellipses_num:
continue
# Discard the ellipses whose center is far away from the center of the innermost ellipse
cluster_center = np.array(candidate_ellipses[0][0])
if max(candidate_ellipses[-1][1]) * scale > 200:
candidate_ellipses = [
e
for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(e[1]) / 5
]
elif max(candidate_ellipses[-1][1]) * scale > 100:
candidate_ellipses = [
e
for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(e[1]) / 10
]
else:
candidate_ellipses = [
e
for e in candidate_ellipses
if LA.norm(e[0] - cluster_center) < max(max(e[1]) / 20, 2)
]
# Discard false positives
if len(candidate_ellipses) < min_ellipses_num:
continue
return [(candidate_ellipses, [[0, 0], [0, 0]])]
return []