在 Python 中使用 OpenCv 中已排序的轮廓对关联的层次结构进行排序
Sort associated Hierarchy with already sorted Contours in OpenCv in Python
我正在使用以下代码从图像中提取最里面的轮廓（input.png）
（我正在使用 Python3.6.3和opencv-python==3.4.0.12)
input.png
import copy
import cv2
BLACK_THRESHOLD = 200
THIN_THRESHOLD = 10
ANNOTATION_COLOUR = (0, 0, 255)
img = cv2.imread('input.png')
orig = copy.copy(img)
gray = cv2.cvtColor(img, 6)
thresh = cv2.threshold(gray, thresh=BLACK_THRESHOLD, maxval=255, type=cv2.THRESH_BINARY_INV)[1]
# Find the contours
_, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_SIMPLE)
hierarchy = hierarchy[0] # get the actual inner list of hierarchy descriptions
idx = 0
# For each contour, find the bounding rectangle and extract it
for component in zip(contours, hierarchy):
currentContour = component[0]
currentHierarchy = component[1]
x, y, w, h = cv2.boundingRect(currentContour)
roi = img[y+2:y + h-2, x+2:x + w-2]
# Skip thin contours (vertical and horizontal lines)
if h < THIN_THRESHOLD or w < THIN_THRESHOLD:
continue
if h > 300 and w > 300:
continue
if h < 40 or w < 40:
continue
if currentHierarchy[3] > 0:
# these are the innermost child components
idx += 1
cv2.imwrite(str(idx) + '.png', roi)
结果：
如您所见，提取的图像没有任何特定顺序。因此，为了解决这个问题，我 根据它们的 x 轴坐标 对轮廓 进行了排序]。下面是代码：
import copy
import cv2
BLACK_THRESHOLD = 200
THIN_THRESHOLD = 10
ANNOTATION_COLOUR = (0, 0, 255)
img = cv2.imread('input.png')
orig = copy.copy(img)
gray = cv2.cvtColor(img, 6)
thresh = cv2.threshold(gray, thresh=BLACK_THRESHOLD, maxval=255, type=cv2.THRESH_BINARY_INV)[1]
# Find the contours
_, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_SIMPLE)
# Sort Contours on the basis of their x-axis coordinates in ascending order
def sort_contours(cnts, method="left-to-right"):
# initialize the reverse flag and sort index
reverse = False
i = 0
# handle if we need to sort in reverse
if method == "right-to-left" or method == "bottom-to-top":
reverse = True
# handle if we are sorting against the y-coordinate rather than
# the x-coordinate of the bounding box
if method == "top-to-bottom" or method == "bottom-to-top":
i = 1
# construct the list of bounding boxes and sort them from top to
# bottom
boundingBoxes = [cv2.boundingRect(c) for c in cnts]
(cnts, boundingBoxes) = zip(*sorted(zip(cnts, boundingBoxes),
key=lambda b: b[1][i], reverse=reverse))
# return the list of sorted contours
return cnts
sorted_contours = sort_contours(contours)
idx = 0
# For each contour, find the bounding rectangle and extract it
for component in sorted_contours:
currentContour = component
x, y, w, h = cv2.boundingRect(currentContour)
roi = img[y + 2:y + h - 2, x + 2:x + w - 2]
# Skip thin contours (vertical and horizontal lines)
if h < THIN_THRESHOLD or w < THIN_THRESHOLD:
continue
if h > 300 and w > 300:
continue
if h < 40 or w < 40:
continue
idx += 1
print(x, idx)
cv2.imwrite(str(idx) + '.png', roi)
结果：
这已经完美地对等高线进行了排序。但是现在你可以看到我得到了所有的轮廓（这是每个数字两份副本的原因）因为我没有使用层次结构 但是当我花了一些时间调试时，我意识到 只有轮廓被排序，而不是它们相关的层次结构 。因此，任何人都可以告诉我如何将层次结构与轮廓一起排序，以便我只能获得已排序轮廓的最内层轮廓。谢谢！
让我们从您的第一个脚本开始，因为它为您提供了不错的结果，只是排序不正确。
观察到基于层次结构的唯一决定（当您决定是否将给定轮廓视为数字时）是 currentHierarchy[3] > 0 为什么我们不从仅选择符合此标准的轮廓开始, 并仅对该子集执行进一步处理（不必再关心层次结构）。
# Find the contours
_, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_SIMPLE)
hierarchy = hierarchy[0] # get the actual inner list of hierarchy descriptions
# Grab only the innermost child components
inner_contours = [c[0] for c in zip(contours, hierarchy) if c[1][3] > 0]
现在我们只剩下我们感兴趣的轮廓，我们只需要对它们进行排序。我们可以重用您原始排序函数的简化版本：
# Sort Contours on the basis of their x-axis coordinates in ascending order
def sort_contours(contours):
# construct the list of bounding boxes and sort them from top to bottom
boundingBoxes = [cv2.boundingRect(c) for c in contours]
(contours, boundingBoxes) = zip(*sorted(zip(contours, boundingBoxes)
, key=lambda b: b[1][0], reverse=False))
# return the list of sorted contours
return contours
并得到排序轮廓：
sorted_contours = sort_contours(inner_contours)
最后，我们要过滤掉垃圾，输出正确标注好的轮廓：
MIN_SIZE = 40
MAX_SIZE = 300
THIN_THRESHOLD = max(10, MIN_SIZE)
PADDING = 2
# ...
idx = 0
# For each contour, find the bounding rectangle and extract it
for contour in sorted_contours:
x, y, w, h = cv2.boundingRect(contour)
roi = img[(y + PADDING):(y + h - PADDING), (x + PADDING):(x + w - PADDING)]
# Skip thin contours (vertical and horizontal lines)
if (h < THIN_THRESHOLD) or (w < THIN_THRESHOLD):
continue
if (h > MAX_SIZE) and (w > MAX_SIZE):
continue
idx += 1
cv2.imwrite(str(idx) + '.png', roi)
完整脚本（使用Python 2.7.x和OpenCV 3.4.1）
import cv2
BLACK_THRESHOLD = 200
MIN_SIZE = 40
MAX_SIZE = 300
THIN_THRESHOLD = max(10, MIN_SIZE)
FILE_NAME = "numbers.png"
PADDING = 2
# ============================================================================
# Sort Contours on the basis of their x-axis coordinates in ascending order
def sort_contours(contours):
# construct the list of bounding boxes and sort them from top to bottom
boundingBoxes = [cv2.boundingRect(c) for c in contours]
(contours, boundingBoxes) = zip(*sorted(zip(contours, boundingBoxes)
, key=lambda b: b[1][0], reverse=False))
# return the list of sorted contours
return contours
# ============================================================================
img = cv2.imread(FILE_NAME)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # Don't use magic numbers
thresh = cv2.threshold(gray, thresh=BLACK_THRESHOLD, maxval=255, type=cv2.THRESH_BINARY_INV)[1]
# Find the contours
_, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_SIMPLE)
hierarchy = hierarchy[0] # get the actual inner list of hierarchy descriptions
# Grab only the innermost child components
inner_contours = [c[0] for c in zip(contours, hierarchy) if c[1][3] > 0]
sorted_contours = sort_contours(inner_contours)
idx = 0
# For each contour, find the bounding rectangle and extract it
for contour in sorted_contours:
x, y, w, h = cv2.boundingRect(contour)
roi = img[(y + PADDING):(y + h - PADDING), (x + PADDING):(x + w - PADDING)]
# Skip thin contours (vertical and horizontal lines)
if (h < THIN_THRESHOLD) or (w < THIN_THRESHOLD):
continue
if (h > MAX_SIZE) and (w > MAX_SIZE):
continue
idx += 1
cv2.imwrite(str(idx) + '.png', roi)
及其生成的图像：
我正在使用以下代码从图像中提取最里面的轮廓（input.png）
（我正在使用 Python3.6.3和opencv-python==3.4.0.12)
input.png
import copy
import cv2
BLACK_THRESHOLD = 200
THIN_THRESHOLD = 10
ANNOTATION_COLOUR = (0, 0, 255)
img = cv2.imread('input.png')
orig = copy.copy(img)
gray = cv2.cvtColor(img, 6)
thresh = cv2.threshold(gray, thresh=BLACK_THRESHOLD, maxval=255, type=cv2.THRESH_BINARY_INV)[1]
# Find the contours
_, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_SIMPLE)
hierarchy = hierarchy[0] # get the actual inner list of hierarchy descriptions
idx = 0
# For each contour, find the bounding rectangle and extract it
for component in zip(contours, hierarchy):
currentContour = component[0]
currentHierarchy = component[1]
x, y, w, h = cv2.boundingRect(currentContour)
roi = img[y+2:y + h-2, x+2:x + w-2]
# Skip thin contours (vertical and horizontal lines)
if h < THIN_THRESHOLD or w < THIN_THRESHOLD:
continue
if h > 300 and w > 300:
continue
if h < 40 or w < 40:
continue
if currentHierarchy[3] > 0:
# these are the innermost child components
idx += 1
cv2.imwrite(str(idx) + '.png', roi)
结果：
如您所见，提取的图像没有任何特定顺序。因此，为了解决这个问题，我 根据它们的 x 轴坐标 对轮廓 进行了排序]。下面是代码：
import copy
import cv2
BLACK_THRESHOLD = 200
THIN_THRESHOLD = 10
ANNOTATION_COLOUR = (0, 0, 255)
img = cv2.imread('input.png')
orig = copy.copy(img)
gray = cv2.cvtColor(img, 6)
thresh = cv2.threshold(gray, thresh=BLACK_THRESHOLD, maxval=255, type=cv2.THRESH_BINARY_INV)[1]
# Find the contours
_, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_SIMPLE)
# Sort Contours on the basis of their x-axis coordinates in ascending order
def sort_contours(cnts, method="left-to-right"):
# initialize the reverse flag and sort index
reverse = False
i = 0
# handle if we need to sort in reverse
if method == "right-to-left" or method == "bottom-to-top":
reverse = True
# handle if we are sorting against the y-coordinate rather than
# the x-coordinate of the bounding box
if method == "top-to-bottom" or method == "bottom-to-top":
i = 1
# construct the list of bounding boxes and sort them from top to
# bottom
boundingBoxes = [cv2.boundingRect(c) for c in cnts]
(cnts, boundingBoxes) = zip(*sorted(zip(cnts, boundingBoxes),
key=lambda b: b[1][i], reverse=reverse))
# return the list of sorted contours
return cnts
sorted_contours = sort_contours(contours)
idx = 0
# For each contour, find the bounding rectangle and extract it
for component in sorted_contours:
currentContour = component
x, y, w, h = cv2.boundingRect(currentContour)
roi = img[y + 2:y + h - 2, x + 2:x + w - 2]
# Skip thin contours (vertical and horizontal lines)
if h < THIN_THRESHOLD or w < THIN_THRESHOLD:
continue
if h > 300 and w > 300:
continue
if h < 40 or w < 40:
continue
idx += 1
print(x, idx)
cv2.imwrite(str(idx) + '.png', roi)
结果：
这已经完美地对等高线进行了排序。但是现在你可以看到我得到了所有的轮廓（这是每个数字两份副本的原因）因为我没有使用层次结构 但是当我花了一些时间调试时，我意识到 只有轮廓被排序，而不是它们相关的层次结构 。因此，任何人都可以告诉我如何将层次结构与轮廓一起排序，以便我只能获得已排序轮廓的最内层轮廓。谢谢！
让我们从您的第一个脚本开始，因为它为您提供了不错的结果，只是排序不正确。
观察到基于层次结构的唯一决定（当您决定是否将给定轮廓视为数字时）是 currentHierarchy[3] > 0 为什么我们不从仅选择符合此标准的轮廓开始, 并仅对该子集执行进一步处理（不必再关心层次结构）。
# Find the contours
_, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_SIMPLE)
hierarchy = hierarchy[0] # get the actual inner list of hierarchy descriptions
# Grab only the innermost child components
inner_contours = [c[0] for c in zip(contours, hierarchy) if c[1][3] > 0]
现在我们只剩下我们感兴趣的轮廓，我们只需要对它们进行排序。我们可以重用您原始排序函数的简化版本：
# Sort Contours on the basis of their x-axis coordinates in ascending order
def sort_contours(contours):
# construct the list of bounding boxes and sort them from top to bottom
boundingBoxes = [cv2.boundingRect(c) for c in contours]
(contours, boundingBoxes) = zip(*sorted(zip(contours, boundingBoxes)
, key=lambda b: b[1][0], reverse=False))
# return the list of sorted contours
return contours
并得到排序轮廓：
sorted_contours = sort_contours(inner_contours)
最后，我们要过滤掉垃圾，输出正确标注好的轮廓：
MIN_SIZE = 40
MAX_SIZE = 300
THIN_THRESHOLD = max(10, MIN_SIZE)
PADDING = 2
# ...
idx = 0
# For each contour, find the bounding rectangle and extract it
for contour in sorted_contours:
x, y, w, h = cv2.boundingRect(contour)
roi = img[(y + PADDING):(y + h - PADDING), (x + PADDING):(x + w - PADDING)]
# Skip thin contours (vertical and horizontal lines)
if (h < THIN_THRESHOLD) or (w < THIN_THRESHOLD):
continue
if (h > MAX_SIZE) and (w > MAX_SIZE):
continue
idx += 1
cv2.imwrite(str(idx) + '.png', roi)
完整脚本（使用Python 2.7.x和OpenCV 3.4.1）
import cv2
BLACK_THRESHOLD = 200
MIN_SIZE = 40
MAX_SIZE = 300
THIN_THRESHOLD = max(10, MIN_SIZE)
FILE_NAME = "numbers.png"
PADDING = 2
# ============================================================================
# Sort Contours on the basis of their x-axis coordinates in ascending order
def sort_contours(contours):
# construct the list of bounding boxes and sort them from top to bottom
boundingBoxes = [cv2.boundingRect(c) for c in contours]
(contours, boundingBoxes) = zip(*sorted(zip(contours, boundingBoxes)
, key=lambda b: b[1][0], reverse=False))
# return the list of sorted contours
return contours
# ============================================================================
img = cv2.imread(FILE_NAME)
gray = cv2.cvtColor(img, cv2.COLOR_BGR2GRAY) # Don't use magic numbers
thresh = cv2.threshold(gray, thresh=BLACK_THRESHOLD, maxval=255, type=cv2.THRESH_BINARY_INV)[1]
# Find the contours
_, contours, hierarchy = cv2.findContours(thresh, cv2.RETR_CCOMP, cv2.CHAIN_APPROX_SIMPLE)
hierarchy = hierarchy[0] # get the actual inner list of hierarchy descriptions
# Grab only the innermost child components
inner_contours = [c[0] for c in zip(contours, hierarchy) if c[1][3] > 0]
sorted_contours = sort_contours(inner_contours)
idx = 0
# For each contour, find the bounding rectangle and extract it
for contour in sorted_contours:
x, y, w, h = cv2.boundingRect(contour)
roi = img[(y + PADDING):(y + h - PADDING), (x + PADDING):(x + w - PADDING)]
# Skip thin contours (vertical and horizontal lines)
if (h < THIN_THRESHOLD) or (w < THIN_THRESHOLD):
continue
if (h > MAX_SIZE) and (w > MAX_SIZE):
continue
idx += 1
cv2.imwrite(str(idx) + '.png', roi)
及其生成的图像：