System, device, and method for pattern representation and recognition
A system and methodologies for pattern representation and recognition are provided. A method includes acquiring a representation associated with discriminating information associated with a subject, retrieving an association between a stored representation and an identity of the subject, determining a discrimination score as a function of the representation and the stored representation based on a neighbor similarity score and relationship contextualization process parameters, and executing one or more control actions based on the discrimination score.
1. A method for pattern representation and recognition, the method comprising:
acquiring, using processing circuitry of a server, a representation associated with discriminating information including a fingerprint sample associated with a subject;
retrieving, using the processing circuitry, an association between a stored representation and an identity of the subject;
determining a neighbor similarity score as a function of distances between two minutiae from two discriminating information representations based on a relationship with a common minutiae identified in the two minutiae;
determining, using the processing circuitry, a discrimination score as a function of the representation and the stored representation based on the neighbor similarity score and whether a contextualizing sub-process was performed on the stored representation, the contextualizing sub-process being performed when two or more operational contexts are used, the two or more operational contexts including different authentication systems or periods of time, the contextualizing sub-process modifying a representation of the fingerprint sample; and
executing, using the processing circuitry, one or more control actions based on the discrimination score.
2. The method of claim 1 , wherein the one or more control actions include:
authenticating the representation of the subject in response to determining that the discrimination score is greater than a predetermined threshold.
3. The method of claim 1 , further comprising:
eliminating an identified selected pairs of neighbor minutiae when paired neighbors do not mutually reciprocate a selection of an original pair as neighbors, each pair include minutiae from the two discriminating information representations.
4. The method of claim 3 , further comprising:
identifying paired sets of minutiae included in the two discriminating information representations by generating a set with a minutia in one discriminating information representation that is not a member of a different set and adding all selected neighbors of all members.