Skip to content

周伟奇 / bmw-ocr

Go to a project
Toggle navigation
Toggle navigation pinning
e2dacd53 by 周伟奇
Options

fix idcard

1 parent 77676d5c
Inline Side-by-side
Showing 3 changed files with 120 additions and 22 deletions
src/apps/doc/consts.py
...@@ -1418,10 +1418,18 @@ ID_COMPARE_LOGIC = { ...@@ -1418,10 +1418,18 @@ ID_COMPARE_LOGIC = {
1418 } 1418 }
1419 1419
1420 PP_COMPARE_LOGIC = { 1420 PP_COMPARE_LOGIC = {
1421 'customerName': ('英文姓名', 'se_name_compare', {'is_passport': True}), 1421 'customerName': ('英文姓名', 'se_name_compare', {'is_passport': True}), # 1
1422 'idNum': ('护照号码', 'se_common_compare', {}), 1422 'idNum': ('护照号码', 'se_common_compare', {}), # 2
1423 'dateOfBirth': ('出生日期', 'se_date_compare', {'input_replace': ''}), 1423 'dateOfBirth': ('出生日期', 'se_date_compare', {'input_replace': ''}), # 2
1424 'idExpiryDate': ('有效期至', 'se_date_compare', {'input_replace': '', 'today': True}), 1424 'idExpiryDate': ('有效期至', 'se_date_compare', {'input_replace': '', 'today': True}), # 2
1425 }
1426
1427 PP_SLICE_MAP = {
1428 'customerChineseName': '机读码1',
1429 'customerName': '机读码1',
1430 'idNum': '机读码2',
1431 'dateOfBirth': '机读码2',
1432 'idExpiryDate': '机读码2',
1425 } 1433 }
1426 1434
1427 EEP_COMPARE_LOGIC = { 1435 EEP_COMPARE_LOGIC = {
...@@ -1582,7 +1590,9 @@ BC_FIELD_ORDER_2 = (('BankName', '发卡行名称'), ...@@ -1582,7 +1590,9 @@ BC_FIELD_ORDER_2 = (('BankName', '发卡行名称'),
1582 IMG_PATH_KEY = 'uniq_img_path_key' 1590 IMG_PATH_KEY = 'uniq_img_path_key'
1583 IMG_PATH_KEY_2 = 'uniq_img_path_key_2' 1591 IMG_PATH_KEY_2 = 'uniq_img_path_key_2'
1584 SECTION_IMG_PATH_KEY = 'uniq_section_img_path_key' 1592 SECTION_IMG_PATH_KEY = 'uniq_section_img_path_key'
1593 SECTION_IMG_PATH_KEY_2 = 'uniq_section_img_path_key_2'
1585 ALL_POSITION_KEY = 'uniq_all_position_key' 1594 ALL_POSITION_KEY = 'uniq_all_position_key'
1595 ALL_POSITION_KEY_2 = 'uniq_all_position_key_2'
1586 POSITION_KEY = 'uniq_position_key' 1596 POSITION_KEY = 'uniq_position_key'
1587 ANGLE_KEY = 'uniq_angle_key' 1597 ANGLE_KEY = 'uniq_angle_key'
1588 FIELD_POSITION_KEY = 'position' 1598 FIELD_POSITION_KEY = 'position'
......
src/apps/doc/management/commands/ocr_process.py
...@@ -305,18 +305,22 @@ class Command(BaseCommand, LoggerMixin): ...@@ -305,18 +305,22 @@ class Command(BaseCommand, LoggerMixin):
305 if not license_data: 305 if not license_data:
306 res_list.append((pno, ino, part_idx, consts.RES_SUCCESS_EMPTY)) 306 res_list.append((pno, ino, part_idx, consts.RES_SUCCESS_EMPTY))
307 return 307 return
308 pre, suf = os.path.splitext(img_path) 308 if isinstance(license_data, dict):
309 base64_img = license_data.pop('base64_img', '') 309 pre, suf = os.path.splitext(img_path)
310 is_save = True if len(base64_img) > 0 else False 310 base64_img = license_data.pop('base64_img', '')
311 section_img_path = '{0}_{1}{2}'.format(pre, part_idx, suf) if is_save else img_path 311 is_save = True if len(base64_img) > 0 else False
312 if is_save: 312 section_img_path = '{0}_{1}{2}'.format(pre, part_idx, suf) if is_save else img_path
313 try: 313 if is_save:
314 with open(section_img_path, "wb") as fh: 314 try:
315 fh.write(base64.b64decode(base64_img.encode())) 315 with open(section_img_path, "wb") as fh:
316 except Exception as e: 316 fh.write(base64.b64decode(base64_img.encode()))
317 self.online_log.warn( 317 except Exception as e:
318 '{0} [section img save failed] [img_path={1}]' 318 self.online_log.warn(
319 ' [part_idx={2}]'.format(self.log_base, img_path, part_idx)) 319 '{0} [section img save failed] [img_path={1}]'
320 ' [part_idx={2}]'.format(self.log_base, img_path, part_idx))
321 else:
322 is_save = False
323 section_img_path = img_path
320 324
321 # 保单 325 # 保单
322 if classify == consts.INSURANCE_CLASSIFY: 326 if classify == consts.INSURANCE_CLASSIFY:
...@@ -517,12 +521,17 @@ class Command(BaseCommand, LoggerMixin): ...@@ -517,12 +521,17 @@ class Command(BaseCommand, LoggerMixin):
517 } 521 }
518 } 522 }
519 523
520 id_card_dict[consts.ALL_POSITION_KEY] = position_dict 524
521 id_card_dict[consts.SECTION_IMG_PATH_KEY] = section_img_path
522 if not is_info_side: 525 if not is_info_side:
523 id_card_dict[consts.IMG_PATH_KEY_2] = img_path 526 id_card_dict[consts.IMG_PATH_KEY_2] = img_path
527 id_card_dict[consts.ALL_POSITION_KEY_2] = position_dict
528 id_card_dict[consts.SECTION_IMG_PATH_KEY_2] = section_img_path
529
524 else: 530 else:
531 id_card_dict[consts.ALL_POSITION_KEY] = position_dict
525 id_card_dict[consts.IMG_PATH_KEY] = img_path 532 id_card_dict[consts.IMG_PATH_KEY] = img_path
533 id_card_dict[consts.SECTION_IMG_PATH_KEY] = section_img_path
534
526 if is_ic and is_save: 535 if is_ic and is_save:
527 card_type = -1 536 card_type = -1
528 json_data_4 = { 537 json_data_4 = {
......
src/celery_compare/tasks.py
...@@ -364,6 +364,7 @@ def ca_compare_license(license_en, ocr_res_dict, field_list): ...@@ -364,6 +364,7 @@ def ca_compare_license(license_en, ocr_res_dict, field_list):
364 ocr_field, compare_logic, special_expiry_date = consts.CA_COMPARE_FIELD[license_en] 364 ocr_field, compare_logic, special_expiry_date = consts.CA_COMPARE_FIELD[license_en]
365 365
366 is_find = False 366 is_find = False
367 special_expiry_date_slice = False
367 result_field_list = [] 368 result_field_list = []
368 section_img_info = dict() 369 section_img_info = dict()
369 field_img_path_dict = dict() 370 field_img_path_dict = dict()
...@@ -423,8 +424,11 @@ def ca_compare_license(license_en, ocr_res_dict, field_list): ...@@ -423,8 +424,11 @@ def ca_compare_license(license_en, ocr_res_dict, field_list):
423 break 424 break
424 425
425 if result == consts.RESULT_N: 426 if result == consts.RESULT_N:
426 next_img_path = expiry_dates_img_path.pop() if len(expiry_dates_img_path) > 0 else empty_str 427 if consts.IMG_PATH_KEY_2 in ocr_res_list[res_idx]:
427 img_path = ocr_res_list[res_idx].get(consts.IMG_PATH_KEY_2, next_img_path) 428 img_path = ocr_res_list[res_idx].get(consts.IMG_PATH_KEY_2, '')
429 special_expiry_date_slice = True
430 else:
431 img_path = expiry_dates_img_path.pop() if len(expiry_dates_img_path) > 0 else empty_str
428 else: 432 else:
429 img_path = empty_str 433 img_path = empty_str
430 else: 434 else:
...@@ -437,12 +441,40 @@ def ca_compare_license(license_en, ocr_res_dict, field_list): ...@@ -437,12 +441,40 @@ def ca_compare_license(license_en, ocr_res_dict, field_list):
437 result_field_list.append((name, value, consts.RESULT_N, empty_str, empty_str, ErrorType.NF.value)) 441 result_field_list.append((name, value, consts.RESULT_N, empty_str, empty_str, ErrorType.NF.value))
438 442
439 if is_find: 443 if is_find:
444 if special_expiry_date_slice:
445 special_section_img_path = section_img_info.get(consts.SECTION_IMG_PATH_KEY_2, '')
446 if os.path.exists(special_section_img_path):
447 field = 'idExpiryDate'
448 special_info = section_img_info.get(consts.ALL_POSITION_KEY_2, {})
449 special_section_position = special_info.get(consts.POSITION_KEY, {})
450 special_section_angle = special_info.get(consts.ANGLE_KEY, 0)
451 try:
452 last_img = img_process(special_section_img_path, special_section_position, special_section_angle)
453 except Exception as e:
454 field_img_path_dict[field] = special_section_img_path
455 else:
456 pre, suf = os.path.splitext(special_section_img_path)
457 try:
458 res_field = compare_logic[field][0]
459 is_valid, coord_tuple = field_build_coordinates(special_info.get(res_field, {}))
460 if is_valid:
461 save_path = '{0}_{1}{2}'.format(pre, field, suf)
462 field_img = last_img[coord_tuple[0]:coord_tuple[1], coord_tuple[2]:coord_tuple[3], :]
463 cv2.imwrite(save_path, field_img)
464 field_img_path_dict[field] = save_path
465 else:
466 field_img_path_dict[field] = special_section_img_path
467 except Exception as e:
468 field_img_path_dict[field] = special_section_img_path
469
440 section_img_path = section_img_info.get(consts.SECTION_IMG_PATH_KEY, '') 470 section_img_path = section_img_info.get(consts.SECTION_IMG_PATH_KEY, '')
441 if os.path.exists(section_img_path): 471 if os.path.exists(section_img_path):
442 failed_field = [] 472 failed_field = []
443 base_img_path = empty_str 473 base_img_path = empty_str
444 for name, _, result, _, img_path, _ in result_field_list: 474 for name, _, result, _, img_path, _ in result_field_list:
445 if result == consts.RESULT_N: 475 if result == consts.RESULT_N:
476 if special_expiry_date_slice and name == 'idExpiryDate':
477 continue
446 failed_field.append(name) 478 failed_field.append(name)
447 if base_img_path == empty_str: 479 if base_img_path == empty_str:
448 base_img_path = img_path 480 base_img_path = img_path
...@@ -459,7 +491,10 @@ def ca_compare_license(license_en, ocr_res_dict, field_list): ...@@ -459,7 +491,10 @@ def ca_compare_license(license_en, ocr_res_dict, field_list):
459 pre, suf = os.path.splitext(section_img_path) 491 pre, suf = os.path.splitext(section_img_path)
460 for field in failed_field: 492 for field in failed_field:
461 try: 493 try:
462 res_field = compare_logic[field][0] 494 if license_en == consts.PP_EN:
495 res_field = consts.PP_SLICE_MAP[field]
496 else:
497 res_field = compare_logic[field][0]
463 is_valid, coord_tuple = field_build_coordinates(info.get(res_field, {})) 498 is_valid, coord_tuple = field_build_coordinates(info.get(res_field, {}))
464 if is_valid: 499 if is_valid:
465 save_path = '{0}_{1}{2}'.format(pre, field, suf) 500 save_path = '{0}_{1}{2}'.format(pre, field, suf)
...@@ -1386,6 +1421,7 @@ def se_compare_license(license_en, ocr_res_dict, field_list): ...@@ -1386,6 +1421,7 @@ def se_compare_license(license_en, ocr_res_dict, field_list):
1386 1421
1387 is_find = False 1422 is_find = False
1388 no_ocr_result = False 1423 no_ocr_result = False
1424 special_expiry_date_slice = False
1389 result_field_list = [] 1425 result_field_list = []
1390 section_img_info = dict() 1426 section_img_info = dict()
1391 field_img_path_dict = dict() 1427 field_img_path_dict = dict()
...@@ -1397,12 +1433,15 @@ def se_compare_license(license_en, ocr_res_dict, field_list): ...@@ -1397,12 +1433,15 @@ def se_compare_license(license_en, ocr_res_dict, field_list):
1397 # 身份证、居住证 过期期限特殊处理 1433 # 身份证、居住证 过期期限特殊处理
1398 if special_expiry_date: 1434 if special_expiry_date:
1399 expiry_dates = set() 1435 expiry_dates = set()
1436 expiry_dates_img_path = set()
1400 key = compare_logic.get('idExpiryDate')[0] 1437 key = compare_logic.get('idExpiryDate')[0]
1401 for ocr_res in ocr_res_list: 1438 for ocr_res in ocr_res_list:
1402 if key in ocr_res: 1439 if key in ocr_res:
1403 expiry_dates.add(ocr_res[key]) 1440 expiry_dates.add(ocr_res[key])
1441 expiry_dates_img_path.add(ocr_res.get(consts.IMG_PATH_KEY_2, ''))
1404 else: 1442 else:
1405 expiry_dates = set() 1443 expiry_dates = set()
1444 expiry_dates_img_path = set()
1406 1445
1407 for res_idx in range(length-1, -1, -1): 1446 for res_idx in range(length-1, -1, -1):
1408 if is_find: 1447 if is_find:
...@@ -1422,6 +1461,9 @@ def se_compare_license(license_en, ocr_res_dict, field_list): ...@@ -1422,6 +1461,9 @@ def se_compare_license(license_en, ocr_res_dict, field_list):
1422 is_find = True 1461 is_find = True
1423 section_img_info[consts.SECTION_IMG_PATH_KEY] = ocr_res_list[res_idx].get(consts.SECTION_IMG_PATH_KEY, '') 1462 section_img_info[consts.SECTION_IMG_PATH_KEY] = ocr_res_list[res_idx].get(consts.SECTION_IMG_PATH_KEY, '')
1424 section_img_info[consts.ALL_POSITION_KEY] = ocr_res_list[res_idx].get(consts.ALL_POSITION_KEY, {}) 1463 section_img_info[consts.ALL_POSITION_KEY] = ocr_res_list[res_idx].get(consts.ALL_POSITION_KEY, {})
1464 if special_expiry_date:
1465 section_img_info[consts.SECTION_IMG_PATH_KEY_2] = ocr_res_list[res_idx].get(
1466 consts.SECTION_IMG_PATH_KEY_2, '')
1425 1467
1426 # 过期期限特殊处理 1468 # 过期期限特殊处理
1427 if special_expiry_date and name == 'idExpiryDate' and result == consts.RESULT_N: 1469 if special_expiry_date and name == 'idExpiryDate' and result == consts.RESULT_N:
...@@ -1432,7 +1474,16 @@ def se_compare_license(license_en, ocr_res_dict, field_list): ...@@ -1432,7 +1474,16 @@ def se_compare_license(license_en, ocr_res_dict, field_list):
1432 result = expiry_date_res 1474 result = expiry_date_res
1433 break 1475 break
1434 1476
1435 img_path = ocr_res_list[res_idx].get(consts.IMG_PATH_KEY, '') if result == consts.RESULT_N else empty_str 1477 if result == consts.RESULT_N:
1478 if consts.IMG_PATH_KEY_2 in ocr_res_list[res_idx]:
1479 img_path = ocr_res_list[res_idx].get(consts.IMG_PATH_KEY_2, '')
1480 special_expiry_date_slice = True
1481 else:
1482 img_path = expiry_dates_img_path.pop() if len(expiry_dates_img_path) > 0 else empty_str
1483 else:
1484 img_path = empty_str
1485 else:
1486 img_path = ocr_res_list[res_idx].get(consts.IMG_PATH_KEY, '') if result == consts.RESULT_N else empty_str
1436 error_type = empty_error_type if result == consts.RESULT_Y else ErrorType.OCR.value 1487 error_type = empty_error_type if result == consts.RESULT_Y else ErrorType.OCR.value
1437 result_field_list.append((name, value, result, ocr_str, img_path, error_type)) 1488 result_field_list.append((name, value, result, ocr_str, img_path, error_type))
1438 else: 1489 else:
...@@ -1443,12 +1494,40 @@ def se_compare_license(license_en, ocr_res_dict, field_list): ...@@ -1443,12 +1494,40 @@ def se_compare_license(license_en, ocr_res_dict, field_list):
1443 result_field_list.append((name, value, consts.RESULT_N, empty_str, empty_str, ErrorType.NF.value)) 1494 result_field_list.append((name, value, consts.RESULT_N, empty_str, empty_str, ErrorType.NF.value))
1444 1495
1445 if is_find: 1496 if is_find:
1497 if special_expiry_date_slice:
1498 special_section_img_path = section_img_info.get(consts.SECTION_IMG_PATH_KEY_2, '')
1499 if os.path.exists(special_section_img_path):
1500 field = 'idExpiryDate'
1501 special_info = section_img_info.get(consts.ALL_POSITION_KEY_2, {})
1502 special_section_position = special_info.get(consts.POSITION_KEY, {})
1503 special_section_angle = special_info.get(consts.ANGLE_KEY, 0)
1504 try:
1505 last_img = img_process(special_section_img_path, special_section_position, special_section_angle)
1506 except Exception as e:
1507 field_img_path_dict[field] = special_section_img_path
1508 else:
1509 pre, suf = os.path.splitext(special_section_img_path)
1510 try:
1511 res_field = compare_logic[field][0]
1512 is_valid, coord_tuple = field_build_coordinates(special_info.get(res_field, {}))
1513 if is_valid:
1514 save_path = '{0}_{1}{2}'.format(pre, field, suf)
1515 field_img = last_img[coord_tuple[0]:coord_tuple[1], coord_tuple[2]:coord_tuple[3], :]
1516 cv2.imwrite(save_path, field_img)
1517 field_img_path_dict[field] = save_path
1518 else:
1519 field_img_path_dict[field] = special_section_img_path
1520 except Exception as e:
1521 field_img_path_dict[field] = special_section_img_path
1522
1446 section_img_path = section_img_info.get(consts.SECTION_IMG_PATH_KEY, '') 1523 section_img_path = section_img_info.get(consts.SECTION_IMG_PATH_KEY, '')
1447 if os.path.exists(section_img_path): 1524 if os.path.exists(section_img_path):
1448 failed_field = [] 1525 failed_field = []
1449 base_img_path = empty_str 1526 base_img_path = empty_str
1450 for name, _, result, _, img_path, _ in result_field_list: 1527 for name, _, result, _, img_path, _ in result_field_list:
1451 if result == consts.RESULT_N: 1528 if result == consts.RESULT_N:
1529 if special_expiry_date_slice and name == 'idExpiryDate':
1530 continue
1452 failed_field.append(name) 1531 failed_field.append(name)
1453 if base_img_path == empty_str: 1532 if base_img_path == empty_str:
1454 base_img_path = img_path 1533 base_img_path = img_path
......
Styling with Markdown is supported
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!