Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-11532178-B2
Publication Date
2022-12-20
Expiration Date
Abstract
In real biometric systems, false match rates and false non-match rates of 0% do not exist. There is always some probability that a purported match is false, and that a genuine match is not identified. The performance of biometric systems is often expressed in part in terms of their false match rate and false non-match rate, with the equal error rate being when the two are equal. There is a tradeoff between the FMR and FNMR in biometric systems which can be adjusted by changing a matching threshold. This matching threshold can be automatically, dynamically and/or user adjusted so that a biometric system of interest can achieve a desired FMR and FNMR.
Core Innovation
The described invention relates to biometric identification and biometric verification in which false match rate (FMR) and false non-match rate (FNMR) are non-zero, so system performance depends on selecting a matching threshold. It generates cumulative histogram data tables for imposter (false match rate) and genuine (false non-match rate) match-score distributions and automatically derives threshold(s) to meet desired FMR/FNMR targets.
The invention further estimates false match probability/confidence using a matchability derived from counts of false-match-score occurrences across comparisons. This matchability is used to produce a false match probability/confidence score to improve match/no match decisions, including more reliable decisions in multi-sample and multi-modal environments.
In addition, the invention supports individualized verification threshold derivation per person using prior genuine/impostor score histories. It applies across one or more biometric-enabled access control, fingerprint search, facial recognition, retinal identification, and iris identification, including scenarios involving multi-sample probes and multi-modal probe samples such as fingerprint and face.
Claims Coverage
The independent claims cover biometric identification and biometric verification in one-to-many gallery search, multi-sample probes with combining information from multiple probe samples, and multi-modal environments that use different modes. Across these independent claims, there is a shared structure that computes match scores, derives false match probability scores/levels from distributions using counts of match scores greater than determined match scores, and then automatically attempts identification to improve match/no match decisions for biometric-enabled applications.
False match probability scoring from gallery and probe match-score distributions for match/no match decisions
Determining a first match score between a biometric probe sample and a gallery sample, determining as the false match probability score a plurality of gallery match scores between the gallery sample and a plurality of other samples, determining a number of the plurality of gallery match scores that are greater than the determined first match score, determining a plurality of probe match scores between the probe sample and the plurality of other samples, determining a number of the plurality of probe match scores that are greater than the determined first match score, and automatically attempting to identify the person by using information from the determining steps to improve match/no match decisions.
Combining multi-sample information to determine a false match probability level for improved identification decisions
Determining a first match score between a first probe sample and a first gallery sample and determining a false match probability score from gallery and probe match-score distributions relative to the first match score, determining a second match score between a second probe sample and a second gallery sample and determining a false match probability score from gallery and probe match-score distributions relative to the second match score, combining information from the first probe sample and the second probe sample to determine a false match probability level, and automatically attempting to identify the person to improve match/no match decisions.
Multi-modal identification using mode-corresponding probe match scores and combined false match probability level
Determining a first match score of a first probe sample corresponding to a mode and a first gallery sample and determining a false match probability score from gallery and probe match-score distributions relative to the first match score, determining a second match score of a second probe sample corresponding to a second mode and a second gallery sample and determining a false match probability score from gallery and probe match-score distributions relative to the second match score, combining information from the first probe sample and the second probe sample to determine a false match probability level, and automatically attempting to identify the person to improve match/no match decisions.
System implementing match-score module to compute false match probability score and attempt identification
Including a match score module with a processor and memory configured to cooperate to determine a first match score between a biometric probe sample and a gallery sample, determine as the false match probability score a plurality of gallery match scores and a number of the plurality of gallery match scores greater than the determined first match score, determine a plurality of probe match scores and a number of the plurality of probe match scores greater than the determined first match score, and automatically attempt to identify the person by using information from the determining steps to improve the match/no match decisions.
System combining first and second probe sample information to determine a false match probability level
Determining a first match score between a first probe sample and a first gallery sample and determining match-score distributions for gallery and probe relative to the first match score, determining a second match score between a second probe sample and a second gallery sample and determining match-score distributions for gallery and probe relative to the second match score, combining information from the first probe sample and the second probe sample to determine a false match probability level, and automatically attempting to identify the person to improve match/no match decisions.
System using first and second mode-corresponding probe match scores to determine false match probability level
Determining a first match score of a first probe sample corresponding to a mode and a first gallery sample and determining match-score distributions for gallery and probe relative to the first match score, determining a second match score of a second probe sample corresponding to a second mode and a second gallery sample and determining match-score distributions for gallery and probe relative to the second match score, combining using the processor information from the first probe sample and the second probe sample to determine a false match probability level, and automatically attempting to identify a person to improve match/no match decisions.
Across the independent claims, the coverage centers on computing match scores for probe and gallery samples, deriving false match probability scores/levels using distributions represented by counts of match scores greater than the determined match scores, and using the resulting information to automatically attempt identification in biometric-enabled access control and biometric modality settings. The independent claims also cover multi-sample combining and multi-modal identification by introducing first and second probe samples corresponding to modes and combining their information into a false match probability level.
Stated Advantages
Improve match/no match decisions.
Automatically attempt to identify the person using information from the determining steps to improve match/no match decisions.
Documented Applications
Biometric-enabled access control.
Fingerprint search.
Facial recognition.
Retinal identification.
Iris identification.
Interested in licensing this patent?