I receive low-quality scans and I need a consistent way to detect when the MRZ is not correct.

Provide a consistent way to validate whether the Machine Readable Zone (MRZ) extracted from low-quality document scans is likely incorrect. This helps reduce downstream errors by flagging unreliable MRZ results for review or re-capture.

This feature enables consistent detection of potentially incorrect MRZ data when source images are low-quality or difficult to read. It evaluates the extracted MRZ against standard MRZ structure expectations, such as required line counts, character set rules, field lengths, and check digit consistency. When the MRZ does not meet expected validity criteria, the result can be flagged as not correct so it can be handled differently in your workflow. This supports scenarios where OCR may produce plausible-looking text that is still invalid due to blur, glare, compression artifacts, or partial occlusion. The primary usage is to run MRZ validation immediately after extraction and before the MRZ is used for identity matching, autofill, or record creation. By detecting invalid MRZ outputs early, you can route cases to manual review, request a better scan, or prompt the user to recapture the document. This helps improve data quality, reduce false matches, and avoid failures later in the onboarding or verification process. It is especially useful for high-volume pipelines that receive heterogeneous scans from different devices and scanning conditions. The outcome is a more reliable, predictable approach to determining whether an MRZ can be trusted, even when the input images are inconsistent.