Limitations of Mobile Device Image Processing

In order to detect if the document is a copy or a real document is included in the authentication term. Authentication is the ability of the software to provide a definitive answer of document authenticity based on document specific security features. The majority of these features can be detected ONLY with UV or IR light.‌

When only WHITE images are uploaded, we can talk about document validation when we can compare data from different data fields (MRZ vs Visual Zone vs Perforation vs Barcode), and some general document layout.

In other words, here it’s important to keep in mind that a user is only using their mobile device to take a capture of an ID document. A mobile device is surely not forensic equipment (unlike the rest of the Regula product line), with no UV, IR, etc optical reader light sources, so it cannot expose all the security elements of a document in order to state its authenticity.

So here we can mostly talk about the WHITE-light image processing with the analysis of certain authenticity elements (where applicable), such as:

• Automatic Document Type detection against the database of supported IDs. The OCR library currently contains over 12.200+ identity document templates
• Verification of MRZ check digits
• Validation of dates for correctness (expiry date, validity, age check etc.)
• Verification of ISO country codes (correctness and existence)
• Validation of document number format for Romanian ID card we have added validation for Romanian CNP number (personal number)
• checkRequiredTextFields - each field in a template will be checked for value presence and if the field is marked as required, but has no value, it will have "error" in validity status.
• Colour check perfectly works for scanned images but not for the photo of the scanned document as it will have color from external light sources
• Cross-comparison and lexical analysis of the obtained information: MRZ <> Viz OCR <> RFID chip to ensure data integrity and consistency