def scan_for_checks(image_filenames: List[str], centers: np.ndarray,
export_dir) -> List[Dict]:
log('Scanning centers for checks in {} files'.format(len(image_filenames)))
page_dicts = []
for page_num, fname in enumerate(image_filenames):
log('Scanning {} ({} of {}):'.format(fname, page_num + 1,
len(image_filenames)))
orig = cv2.imread(fname)
orig_height, _, _ = orig.shape
resize_factor = wanted_height / orig_height
resize = cv2.resize(orig, (0, 0), fx=resize_factor, fy=resize_factor)
grey = cv2.cvtColor(resize, cv2.COLOR_RGB2GRAY)
thresh = cv2.adaptiveThreshold(grey, 255, cv2.ADAPTIVE_THRESH_MEAN_C,
cv2.THRESH_BINARY, 11, 2)
erode_kernel = np.ones((3, 3), np.uint8)
erode = cv2.erode(thresh, erode_kernel)
boxes_dict = {}
for center_num, center in enumerate(centers):
x, y = center[0], center[1]
mean = np.mean(
erode[int(y - check_range / 2):int(y + check_range / 2),
int(x - check_range / 2):int(x + check_range / 2)])
cv2.circle(resize, (x, y), 2, color_blue, thickness=2)
if mean < blank_box_mean - check_thresh:
boxes_dict[center_num] = True
cv2.circle(resize, (x, y),
circle_radius,
color_green,
thickness=3)
page_dicts.append({'page': page_num, 'boxes': boxes_dict})
cv2.imwrite('{}/{}.jpg'.format(export_dir, str(page_num)), resize)
return page_dicts
def get_args() -> Tuple[str, int]:
pdf_fname = ""
k = 0
try:
pdf_fname = sys.argv[1]
k = int(sys.argv[2])
except (ValueError, IndexError):
log("Usage: ./main.py file.pdf k")
exit(1)
finally:
pass
if not path.isfile(pdf_fname):
log("{} is not a valid file".format(pdf_fname))
exit(1)
return pdf_fname, k
def convert_to_jpg(pdf_filename, export_path) -> list:
# Convert to jpg
log('Starting conversion to jpg...')
start = timeit.default_timer()
pages = convert_from_path(pdf_filename, 500)
end = timeit.default_timer()
log('Successfully converted %d pages in %d seconds.' %
(len(pages), end - start))
# Save images
log('Saving to %s/...' % export_path)
image_filenames = []
for i, page in enumerate(pages):
image_filename = '%s/p%d.jpg' % (export_path, i)
image_filenames.append(image_filename)
page.save(image_filename, 'JPEG')
log('Saved all images')
return image_filenames
def scan_for_squares(image_filenames) -> np.ndarray:
log('Looking for checkboxes in {} images'.format(len(image_filenames)))
centers = []
for i, fname in enumerate(image_filenames):
log('Scanning {} ({} of {}):'.format(fname, i + 1,
len(image_filenames)),
end=' ')
orig = cv2.imread(fname)
orig_height, _, _ = orig.shape
resize_factor = wanted_height / orig_height
resize = cv2.resize(orig, (0, 0), fx=resize_factor, fy=resize_factor)
img2 = cv2.cvtColor(resize, cv2.COLOR_RGB2GRAY)
canny = cv2.Canny(img2, 0, 100)
dilate_kernel = np.ones((5, 5), np.uint8)
dilate = cv2.dilate(canny, dilate_kernel)
erode_kernel = np.ones((5, 5), np.uint8)
erode = cv2.erode(dilate, erode_kernel)
contours = cv2.findContours(erode.copy(), cv2.RETR_EXTERNAL,
cv2.CHAIN_APPROX_SIMPLE)
contours = imutils.grab_contours(contours)
center_count = 0
for c in contours:
peri = cv2.arcLength(c, True)
c_approx = cv2.approxPolyDP(c, 0.04 * peri, True)
if len(c_approx) == 4 \
and square_area - area_thresh < cv2.contourArea(c) < square_area + area_thresh \
and square_perimeter - perimeter_thresh < peri < square_perimeter + perimeter_thresh:
cv2.drawContours(resize, [c_approx], -1, color_green, 4)
moments = cv2.moments(c)
center_x = int(moments["m10"] / moments["m00"])
center_y = int(moments["m01"] / moments["m00"])
centers.append([center_x, center_y])
center_count += 1
log('{} boxes'.format(center_count))
if len(centers) == 0:
return np.empty(0) # Cannot np.vstack empty list
else:
return np.vstack(centers)
except (ValueError, IndexError):
log("Usage: ./main.py file.pdf k")
exit(1)
finally:
pass
if not path.isfile(pdf_fname):
log("{} is not a valid file".format(pdf_fname))
exit(1)
return pdf_fname, k
(pdf_filename, k_centers) = get_args()
log("Starting survey scan of {}. Looking for {} boxes.".format(
pdf_filename, k_centers))
export_dirname = '/tmp/survey/{}'.format(uuid.uuid4())
makedirs(export_dirname)
raw_dir = '{}/raw'.format(export_dirname)
makedirs(raw_dir)
test_image_filenames = convert_to_jpg(pdf_filename, raw_dir)
log('Starting scan for boxes.')
start = timeit.default_timer()
test_centers = scan_for_squares(test_image_filenames)
end = timeit.default_timer()
log('Scanned {} pages in {} seconds.'.format(len(test_image_filenames),
int(end - start)))