omnimrz 0.2.0

Fast and robust MRZ extraction, parsing, and validation using PaddleOCR

OmniMRZ — Python MRZ Extraction & Validation Library for Passport OCR and KYC

OmniMRZ is an open-source Python library for Machine Readable Zone (MRZ) extraction, parsing, and ICAO-9303 validation from passport and ID images, built for OCR, KYC, and identity verification systems.

It is a production-grade MRZ extraction and validation engine designed for high-accuracy KYC, identity verification, and document intelligence pipelines.

Unlike simple MRZ readers, OmniMRZ evaluates whether an MRZ is structurally correct, cryptographically valid, and logically plausible.

Typical Use Cases

🛂 Passport and ID card OCR pipelines
🏦 KYC / AML identity verification systems
✈️ Border control and immigration preprocessing
📄 Document digitization and archiving
🔐 Authentication and onboarding workflows

⭐ Show Your Support If OmniMRZ helped you or saved development time: 👉 Please consider starring the repository It helps visibility and motivates continued development

Features

Installation

Contributing

Why OmniMRZ?

Unlike basic MRZ readers, OmniMRZ provides end-to-end MRZ quality assurance:

• Combines OCR, structural validation, checksum verification, and logical consistency checks
• Fully compliant with ICAO 9303
• Designed for production KYC and identity verification systems
• Robust against OCR noise and partially corrupted MRZ lines

Features

At a glance

• MRZ detection and extraction from images
• Supports TD3 (passport) format
• Checksum validation (ICAO 9303)
• Logical and structural validation
• Clean Python API

Detailed features

🔍 MRZ Extraction

• PaddleOCR-based MRZ text extraction (robust on mobile & noisy images)
• Intelligent MRZ line clustering & reconstruction
• Automatic MRZ type detection (TD1 / TD2 / TD3)
• OCR noise filtering & MRZ-safe character normalization
• Works even with partially corrupted or misaligned MRZs

🧱 Structural Validation (ICAO 9303)

• Exact line-length enforcement
• Strict MRZ format verification
• Field-level structural checks
• Early-exit gating for invalid layouts

🔢 Checksum Validation

• Fully ICAO-9303 compliant checksum algorithm
• Field-level validation:
• Document number
• Date of birth
• Expiry date
• Composite checksum
• OCR-error tolerant digit correction (O→0, S→5, B→8, etc.)
• Detailed checksum failure diagnostics

🧠 Logical & Semantic Validation

• Expired document detection
• Future date-of-birth detection
• Implausible age detection
• DOB ≥ expiry detection
• Gender value validation (M, F, X, <)
• Cross-field consistency signals (issuer vs nationality)

📤 Output

• Clean MRZ text
• Structured JSON
• Deterministic pass / fail / warning signals
• Human-readable error messages

Installation

pip install omnimrz

Note: PaddleOCR requires additional system dependencies. Please ensure PaddlePaddle installs correctly on your platform.

pip install paddleocr
pip install paddle paddle

or if that fails then run

python -m pip install paddlepaddle==3.0.0 -i https://www.paddlepaddle.org.cn/packages/stable/cpu/

Quick Usage

from omnimrz import OmniMRZ

omni = OmniMRZ()
result = omni.process("ukpassport.jpg")

print(result)

Output Example

{
    "extraction": {
        "status": "SUCCESS(extraction of mrz)",
        "line1": "P<GBRPUDARSAN<<HENERT<<<<<<<<<<<<<<<<<<<<<<<",
        "line2": "7077979792GBR9505209M1704224<<<<<<<<<<<<<<00" 
    },
    "structural_validation": {
        "status": "PASS",
        "mrz_type": "TD3",
        "errors": []
    },
    "checksum_validation": {
        "status": "PASS",
        "errors": []
    },
    "parsed_data": {
        "status": "PARSED",
        "data": {
            "document_type": "P",
            "issuing_country": "GBR",
            "surname": "PUDARSAN",
            "given_names": "HENERT",
            "document_number": "707797979",
            "nationality": "GBR",
            "date_of_birth": "1995-05-20",
            "gender": "M",
            "expiry_date": "2017-04-22",
            "personal_number": ""
        }
    },
    "logical_validation": {
        "status": "FAIL",
        "errors": [
            "DOCUMENT_EXPIRED"
        ]
    },
    "screenshot_detection": {
        "status": "PASS",
        "is_screenshot": false,
        "score": 3,
        "confidence": 30.0,
        "reasons": [
            "Low ELA: 0.38",
            "High horizontal edges: 0.51",
            "High sharpness: 2029.58"
        ]
    }
}

Citing OmniMRZ

If you use OmniMRZ in academic research or publications, please consider citing this repository:

Contributing

Contributions are welcome!🤝

1. Fork the repository
2. Create your feature branch

git checkout -b feature/amazing-feature

3. Commit your changes
4. Push to your branch
5. Open a Pull Request

Keywords

MRZ extraction, passport OCR, machine readable zone, ICAO 9303, MRZ parser, Python OCR, identity verification, KYC automation, document intelligence, ID card scanning, border control OCR

misc