Score for whole user record (consolidated score) is the maximum score for individual fingerprint pairs minus some constant correction. This constant depends on number of fingerprints compared.

The correction constant is intended to keep FAR stable for given threshold no matter how many fingerprints are used for matching. Without this correction, multi-finger user records would statistically receive higher score that single fingerprint user records, because choice of maximum score tends to accumulate random variations in matching score.


In IDKit SDK version 2.70 and newer the consolidated score is calculated as the sum of scores of each corresponding fingerprints pair. Situation is more complicated if user record contains multiple fingerprint template instances of same fingerprint position (e.g., user record contains three right index fingerprint templates in order to achieve better accuracy). All similarity scores of corresponding fingerprint pairs are put into set based on fingerprint position. Maximum similarity score is then evaluated for each fingerprint position set and consolidated score is summation of all those maximums.



Example

Probe user record A contains: 

Right Index fingerprint template (index 0) 

Left Index fingerprint template (index 1) 

Gallery user record B contains: 

Right Index fingerprint template (index 0) 

Right Index fingerprint template (index 1) 

Left Index fingerprint template (index 2) 

 

Similarity scores evaluated by verification algorithm: 

SCORE A.0_B.0 

SCORE A.0_B.1 

SCORE A.1_B.2 

 

Consolidated score = MAX(SCORE A.0_B.0, SCORE A.0_B.1) + SCORE A.1_B.2