MRZ reading¶
MRZ text filling validity is determined in accordance with ICAO 9303 for machine readable documents. Unlike document reading, it doesn’t not require a previous detection of the document.
MRZ reading module provide functions to read the MRZ as the full string, and to decode its field as a dictionary name:value.
Fields names by MRZ Types¶
TD2, TD3, MRVA & B |
TD1 |
French ID |
Driving License |
|
|---|---|---|---|---|
primary_name |
✔️ |
✔️ |
✔️ |
|
secondary_name |
✔️ |
✔️ |
✔️ |
|
type |
✔️ |
✔️ |
✔️ |
|
issuing_state |
✔️ |
✔️ |
✔️ |
|
document_number |
✔️ |
✔️ |
✔️ |
|
nationality |
✔️ |
✔️ |
||
birth_date |
✔️ |
✔️ |
✔️ |
|
gender |
✔️ |
✔️ |
✔️ |
|
expiry_date |
✔️ |
✔️ |
||
personal_number |
✔️ |
|||
optional_1 |
✔️ |
|||
optional_2 |
✔️ |
|||
check_digit_1 |
✔️ |
✔️ |
✔️ |
|
check_digit_2 |
✔️ |
✔️ |
✔️ |
|
check_digit_3 |
✔️ |
✔️ |
✔️ |
|
check_digit_4 |
✔️ |
✔️ |
||
check_digit_5 |
✔️ |
|||
issuing_department |
✔️ |
|||
issuing_office |
✔️ |
|||
issuing_department_2 |
✔️ |
|||
issuing_date |
✔️ |
Example¶
The example below demonstrates how to read, check and decode an MRZ from an image.
# load the deep learning model
id3document.DocumentLibrary.load_model(ai_models_path, id3document.DocumentModel.MRZ_READER_2A, id3document.ProcessingUnit.CPU)
# initialize the MRZ reader
mrz_reader = id3document.MrzReader(
thread_count=4
)
# load an image from a file.
image = id3document.DocumentImage.from_file(filename, id3document.PixelFormat.BGR_24_BITS)
# read the MRZ
mrz_result = mrz_reader.read_mrz(image)
print(f"Confidence : {mrz_result.confidence}")
print(f"Value : {mrz_result.mrz}")
print(f"Type : {mrz_result.mrz_type.name}")
# check the validity of the MRZ
mrz_valid = id3document.MrzHelper.check(mrz_result.mrz, mrz_result.mrz_type)
# decode the MRZ
mrz_fields = id3document.MrzHelper.decode(mrz_result.mrz, mrz_result.mrz_type)
for field in mrz_fields:
print(f"- {field} = {mrz_fields[field]}")