27 lines
842 B
Python
27 lines
842 B
Python
from numpy import array, ndarray
|
|
import pdf2image
|
|
import cv2
|
|
|
|
from vidocp.utils.deskew import deskew
|
|
|
|
|
|
def preprocess_pdf_image(page):
|
|
if len(page.shape) > 2:
|
|
page = cv2.cvtColor(page, cv2.COLOR_BGR2GRAY)
|
|
page = cv2.fastNlMeansDenoising(page, h=3)
|
|
return deskew(page)
|
|
|
|
|
|
def open_pdf(pdf, first_page=0, last_page=None):
|
|
first_page += 1
|
|
last_page = None if last_page is None else last_page + 1
|
|
if type(pdf) == str:
|
|
pages = pdf2image.convert_from_path(pdf, first_page=first_page, last_page=last_page)
|
|
elif type(pdf) == bytes:
|
|
pages = pdf2image.convert_from_bytes(pdf, first_page=first_page, last_page=last_page)
|
|
elif type(pdf) in {list, ndarray}:
|
|
return pdf
|
|
pages = [preprocess_pdf_image(array(p)) for p in pages]
|
|
pages, angles = list(zip(*pages))
|
|
return pages, angles
|