How accurate is MRZ scanning?

MRZ scanning, when conditions are reasonable, tends to be one of the more accurate parts of the identity document extraction process, largely because of the standardization discussed elsewhere in this FAQ: a fixed font, a fixed field layout, and a restricted character set. Vendors across the identity verification industry commonly report accuracy figures in the high 90s or above for MRZ extraction under good capture conditions, though the exact number depends heavily on image quality, document condition, and how the figure is measured.

A few factors drive accuracy down in practice. A worn or faded MRZ, common on older or heavily used passports, is harder to read reliably.

Poor lighting, glare, or a document held at a sharp angle during capture introduces the same kinds of errors any OCR system faces. Low-resolution cameras or heavily compressed images can blur character edges just enough to cause misreads, particularly between visually similar characters.

Good scanning software addresses most of this at the capture stage rather than trying to fix it after the fact, giving users real-time feedback to hold the document flatter, move closer, or find better lighting before the image is even processed, since a clean capture upstream avoids most of the downstream extraction problems. Check-digit validation then catches many of the errors that do slip through, since a misread character will usually cause the recalculated check digit to fail to match the printed one, flagging the field for review rather than passing along incorrect data silently.

It's worth being precise about what "accuracy" means here: it usually refers to correctly extracted fields per document scanned under specified conditions, not a claim that every single scan under every possible condition will be flawless. Reasonable due diligence when evaluating any vendor's accuracy claims includes asking what test conditions and document set were used to generate the number.

ScanDoc reports data extraction accuracy in the high-90s to near-100% range on its product pages, achieved through the combination of check-digit validation, cross-referencing against visual zone OCR, and capture-stage guidance that reduces the number of poor-quality images reaching the OCR engine in the first place.