def combine_boxes(boxes, img_dim, dist_threshold=15, padding=0):
"""Uses UnionFind to group close-by contours into boxes (disjoint
connected components).
Parameters:
boxes (list(numpy array)): List of numpy arrays of coordinates of a
box as returned by enclosing_box.
img_dim (tuple(int)): Tuple of image dimensions.
dis_threshold (int): Threshold of number of pixels to determine
whether boxes are too close.
padding (int): Number of pixels that are padding.
Return:
(numpy array): Numpy array of combined boxes.
"""
n = len(boxes)
uf = UnionFind(n)
for i, j in combinations(range(n), 2):
if closest_distance(boxes[i], boxes[j]) < dist_threshold:
uf.union(i, j)
box_groups = [[box for i in group for box in boxes[i]]
for group in uf.groups()]
combined_boxes = [
enclosing_box(group, img_dim, padding=padding) for group in box_groups
]
filtered_boxes = [x for x in combined_boxes if fits_criteria_box(x)]
return np.array(filtered_boxes)
def accounts_merge_using_union_find(
accounts: List[List[str]]) -> List[List[str]]:
uf = UnionFind()
email_to_name = dict()
for account in accounts:
name = account[0]
for email in account[1:]:
email_to_name[email] = name
uf.union(email, account[1])
return [[email_to_name[pmail]] + sorted(emails)
for (pmail, emails) in uf.groups().items()]
def combine_boxes(boxes, img_dim, dist_threshold=15, padding=0):
'''
Uses UnionFind to group close-by contours into boxes (disjoint
connected components), for example, to combine letters into a text box
'''
n = len(boxes)
uf = UnionFind(n)
for i, j in combinations(range(n), 2):
if closest_distance(boxes[i], boxes[j]) < dist_threshold:
uf.union(i, j)
box_groups = [[box for i in group for box in boxes[i]]
for group in uf.groups()]
combined_boxes = [
enclosing_box(group, img_dim, padding=padding) for group in box_groups
]
filtered_boxes = [x for x in combined_boxes if fits_criteria_box(x)]
return np.array(filtered_boxes)