123 lines
4.4 KiB
Python
123 lines
4.4 KiB
Python
from itertools import count
|
|
|
|
import cv2
|
|
import numpy as np
|
|
import pdf2image
|
|
from matplotlib import pyplot as plt
|
|
from timeit import timeit
|
|
|
|
|
|
def parse(image: np.array):
|
|
gray_scale = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
|
|
th1, img_bin = cv2.threshold(gray_scale, 200, 255, cv2.THRESH_BINARY)
|
|
img_bin = ~img_bin
|
|
|
|
line_min_width = 5
|
|
kernel_h = np.ones((1, line_min_width), np.uint8)
|
|
kernel_v = np.ones((line_min_width, 1), np.uint8)
|
|
|
|
img_bin_h = cv2.morphologyEx(img_bin, cv2.MORPH_OPEN, kernel_h)
|
|
img_bin_v = cv2.morphologyEx(img_bin, cv2.MORPH_OPEN, kernel_v)
|
|
# print([cv2.countNonZero(row) for row in img_bin_v])
|
|
|
|
img_bin_final = img_bin_h | img_bin_v
|
|
|
|
find_and_close_edges(img_bin_final)
|
|
|
|
_, labels, stats, _ = cv2.connectedComponentsWithStats(~img_bin_final, connectivity=8, ltype=cv2.CV_32S)
|
|
return labels, stats
|
|
|
|
|
|
# def filter_unconnected_cells(stats):
|
|
# filtered_cells = []
|
|
# for left, middle, right in zip(stats[0:], stats[1:], list(stats[2:])+[None]):
|
|
# x, y, w, h, area = middle
|
|
# if w > 35 and h > 13 and area > 500:
|
|
# if y == left[1] or y == right[1]:
|
|
# filtered_cells.append(middle)
|
|
# return filtered_cells
|
|
|
|
def filter_unconnected_cells(stats):
|
|
filtered_cells = []
|
|
# print(stats)
|
|
for left, middle, right in zip(stats[0:], stats[1:], list(stats[2:]) + [np.array([None, None, None, None, None])]):
|
|
x, y, w, h, area = middle
|
|
if w > 35 and h > 13 and area > 500:
|
|
if right[1] is None:
|
|
if y == left[1] or x == left[0]:
|
|
filtered_cells.append(middle)
|
|
else:
|
|
if y == left[1] or y == right[1] or x == left[0] or x == right[0]:
|
|
filtered_cells.append(middle)
|
|
return filtered_cells
|
|
|
|
|
|
# def annotate_image(image, stats):
|
|
# stats = filter_unconnected_cells(stats)
|
|
# for i,val in enumerate(stats):
|
|
# x, y, w, h, area = stats[i][0], stats[i][1], stats[i][2], stats[i][3], stats[i][4]
|
|
# cv2.rectangle(image, (x, y), (x + w, y + h), (255, 0, 255), 2)
|
|
#
|
|
# for i, (s, v) in enumerate(zip(["x", "y", "w", "h"], [x, y, w, h])):
|
|
# anno = f"{s} = {v}"
|
|
# xann = int(x + 5)
|
|
# yann = int(y + h - (20 * (i + 1)))
|
|
# cv2.putText(image, anno, (xann, yann), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 0, 255), 2)
|
|
#
|
|
# return image
|
|
def annotate_image(image, stats):
|
|
stats = filter_unconnected_cells(stats)
|
|
for stat in stats:
|
|
x, y, w, h, area = stat
|
|
cv2.rectangle(image, (x, y), (x + w, y + h), (255, 0, 255), 2)
|
|
for i, (s, v) in enumerate(zip(["x", "y", "w", "h"], [x, y, w, h])):
|
|
anno = f"{s} = {v}"
|
|
xann = int(x + 5)
|
|
yann = int(y + h - (20 * (i + 1)))
|
|
cv2.putText(image, anno, (xann, yann), cv2.FONT_HERSHEY_SIMPLEX, 0.7, (255, 0, 255), 2)
|
|
|
|
return image
|
|
|
|
|
|
def find_and_close_edges(img_bin_final):
|
|
contours, hierarchy = cv2.findContours(img_bin_final, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
|
|
for cnt in contours:
|
|
missing_external_edges = True
|
|
left = tuple(cnt[cnt[:, :, 0].argmin()][0])
|
|
right = tuple(cnt[cnt[:, :, 0].argmax()][0])
|
|
top = tuple(cnt[cnt[:, :, 1].argmin()][0])
|
|
bottom = tuple(cnt[cnt[:, :, 1].argmax()][0])
|
|
topleft = [left[0] + 1, top[1]]
|
|
# print(cnt, left, top, topleft)
|
|
bottomright = [right[0] - 1, bottom[1]]
|
|
for arr in cnt:
|
|
if np.array_equal(arr, np.array([bottomright])) or np.array_equal(arr, np.array([topleft])):
|
|
missing_external_edges = False
|
|
break
|
|
|
|
if missing_external_edges and (bottomright[0]-topleft[0])*(bottomright[1]-topleft[1])>= 50000:
|
|
topleft[0] -= 1
|
|
bottomright[0] += 1
|
|
cv2.rectangle(img_bin_final, tuple(topleft), tuple(bottomright), (255,255,255) , 2)
|
|
print("missing cell detectet rectangle drawn")
|
|
|
|
return img_bin_final
|
|
|
|
|
|
def parse_tables_in_pdf(pages):
|
|
return zip(map(parse, pages), count())
|
|
|
|
|
|
def annotate_tables_in_pdf(pdf_path, page_index=1):
|
|
timeit()
|
|
page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
|
|
page = np.array(page)
|
|
|
|
_, stats = parse(page)
|
|
page = annotate_image(page, stats)
|
|
print(timeit())
|
|
fig, ax = plt.subplots(1, 1)
|
|
fig.set_size_inches(20, 20)
|
|
ax.imshow(page)
|
|
plt.show()
|