140 lines
4.1 KiB
Python
140 lines
4.1 KiB
Python
from itertools import compress
|
|
from itertools import starmap
|
|
from operator import __and__
|
|
|
|
import cv2
|
|
import numpy as np
|
|
from pdf2image import pdf2image
|
|
|
|
from cv_analysis.utils.display import show_mpl
|
|
from cv_analysis.utils.draw import draw_rectangles
|
|
from cv_analysis.utils.post_processing import remove_overlapping, remove_included, has_no_parent
|
|
from cv_analysis.utils.visual_logging import vizlogger
|
|
|
|
|
|
def is_likely_segment(rect, min_area=100):
|
|
return cv2.contourArea(rect, False) > min_area
|
|
|
|
|
|
def find_segments(image):
|
|
contours, hierarchies = cv2.findContours(image, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
|
|
|
|
mask1 = map(is_likely_segment, contours)
|
|
mask2 = map(has_no_parent, hierarchies[0])
|
|
mask = starmap(__and__, zip(mask1, mask2))
|
|
contours = compress(contours, mask)
|
|
|
|
rectangles = (cv2.boundingRect(c) for c in contours)
|
|
|
|
return rectangles
|
|
|
|
|
|
def parse_layout(image: np.array):
|
|
|
|
image = image.copy()
|
|
image_ = image.copy()
|
|
|
|
if len(image_.shape) > 2:
|
|
image_ = cv2.cvtColor(image_, cv2.COLOR_BGR2GRAY)
|
|
vizlogger.debug(image_, "layout01_start.png")
|
|
|
|
image_ = cv2.GaussianBlur(image_, (7, 7), 0)
|
|
vizlogger.debug(image_, "layout02_blur.png")
|
|
thresh = cv2.threshold(image_, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]
|
|
vizlogger.debug(image_, "layout03_theshold.png")
|
|
|
|
kernel = cv2.getStructuringElement(cv2.MORPH_RECT, (5, 5))
|
|
vizlogger.debug(kernel, "layout04_kernel.png")
|
|
dilate = cv2.dilate(thresh, kernel, iterations=4)
|
|
vizlogger.debug(dilate, "layout05_dilate.png")
|
|
|
|
rects = list(find_segments(dilate))
|
|
|
|
# -> Run meta detection on the previous detections TODO: refactor
|
|
for rect in rects:
|
|
x, y, w, h = rect
|
|
cv2.rectangle(image, (x, y), (x + w, y + h), (0, 0, 0), -1)
|
|
cv2.rectangle(image, (x, y), (x + w, y + h), (255, 255, 255), 7)
|
|
vizlogger.debug(image, "layout06_rectangles.png")
|
|
|
|
_, image = cv2.threshold(image, 254, 255, cv2.THRESH_BINARY)
|
|
vizlogger.debug(image, "layout07_threshold.png")
|
|
image = ~image
|
|
vizlogger.debug(image, "layout08_inverse.png")
|
|
|
|
if len(image.shape) > 2:
|
|
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
|
|
vizlogger.debug(image, "layout09_convertcolor.png")
|
|
|
|
rects = find_segments(image)
|
|
# <- End of meta detection
|
|
|
|
rects = remove_included(rects)
|
|
rects = remove_overlapping(rects)
|
|
|
|
return list(rects)
|
|
|
|
|
|
def annotate_layout_in_pdf(pdf_path, page_index=1, show=False):
|
|
|
|
page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
|
|
page = np.array(page)
|
|
|
|
rects = parse_layout(page)
|
|
page = draw_rectangles(page, rects)
|
|
vizlogger.debug(page, "layout10_output.png")
|
|
|
|
if show:
|
|
show_mpl(page)
|
|
|
|
|
|
"""
|
|
def find_layout_boxes(image: np.array):
|
|
|
|
if len(image.shape) > 2:
|
|
image = cv2.cvtColor(image, cv2.COLOR_BGR2GRAY)
|
|
image = cv2.GaussianBlur(image, (5, 5), 1)
|
|
image = cv2.threshold(image, 253, 255, cv2.THRESH_BINARY)[1]
|
|
img_bin = ~image
|
|
|
|
line_min_width = 10
|
|
kernel_h = np.ones((10, line_min_width), np.uint8)
|
|
kernel_v = np.ones((line_min_width, 10), 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)
|
|
|
|
img_bin_final = img_bin_h | img_bin_v
|
|
|
|
contours = cv2.findContours(img_bin_final, cv2.RETR_EXTERNAL, cv2.CHAIN_APPROX_SIMPLE)
|
|
contours = imutils.grab_contours(contours)
|
|
|
|
for c in contours:
|
|
peri = cv2.arcLength(c, True)
|
|
approx = cv2.approxPolyDP(c, 0.04 * peri, True)
|
|
yield cv2.boundingRect(approx)
|
|
|
|
|
|
def annotate_layout_boxes(image, rects):
|
|
for rect in rects:
|
|
(x, y, w, h) = rect
|
|
cv2.rectangle(image, (x, y), (x + w, y + h), (0, 255, 0), 2)
|
|
|
|
return image
|
|
|
|
|
|
def annotate_layout_in_pdf(pdf_path, page_index=1):
|
|
|
|
page = pdf2image.convert_from_path(pdf_path, first_page=page_index + 1, last_page=page_index + 1)[0]
|
|
page = np.array(page)
|
|
|
|
layout_boxes = find_layout_boxes(page)
|
|
page = annotate_layout_boxes(page, layout_boxes)
|
|
|
|
fig, ax = plt.subplots(1, 1)
|
|
fig.set_size_inches(20, 20)
|
|
ax.imshow(page)
|
|
plt.show()
|
|
|
|
"""
|