Sensor dependency
Fingerprint quality depends on the sensor and capture mode. Contact, contactless, single-finger, slap, rolled and latent are different profiles; Current documentation claims none without a source/test matrix.
The technical chain covers capture, quality, feature/minutiae concepts, enrollment, 1:1 verification, 1:N identification and duplicate search without inventing proprietary algorithms or benchmark values.
Sensor acquisition → segmentation/quality → enrollment/template → 1:1/1:N → evidence
Fingerprint quality depends on the sensor and capture mode. Contact, contactless, single-finger, slap, rolled and latent are different profiles; Current documentation claims none without a source/test matrix.
Where capture contains multiple finger regions, segmentation and quality handling precede enrollment/matching. Numeric quality requirements are published only with engine/test evidence.
Enrollment binds samples to a subject and device/process provenance. Duplicate or low-quality enrollment needs explicit failure handling; template format/version is not invented.
1:1 verifies a claimed subject/reference and is distinct from identification. Threshold and FAR/FMR operating points require benchmark context.
1:N and deduplication require gallery definition, candidate handling, no-match/ambiguous semantics and adjudication where supported. API group initialization proves the identification-group concept, not a complete ABIS workflow.
Evidence should separate capture/source metadata, quality/result, matcher outcome and policy decision. Sensor serial/firmware appears only when a device test matrix permits publication.