Article 36: This article focuses on multi-biometric data acquisition and processing architecture. The human computer interface used by a multi-biometric system impacts its usability. In particular, the sequence of biometric data acquisition impacts the user’s convenience. A multi-biometric system must be designed in such a way that the enrolment time and the Failure To Enrol (FTE) rate are significantly reduced by designing an acquisition protocol that enhances the user’s convenience while ensuring that good quality biometric data is obtained from the user. In addition, the sequence in which the captured biometric data is processed can massively impact the throughput time in large-scale identification systems.
The designer must seek methods that optimise the multi-biometric architecture by minimising the enrolment time and FTE in order to arrive at an identification decision rapidly. The various types of acquisition and processing architectures are outlined below:
Multi-biometric Acquisition sequence – This refers to the order in which the different sources of biometric evidence are acquired from an individual. It is worth highlighting that a multi-algorithm biometric system, see previous articles for details, uses only a single biometric sample. Therefore, the acquisition methodology is not a concern. With multi-biometric systems which employees more than one biometric sample, the evidence is gathered sequentially as each source is independently obtained with a short time interval between successive acquisitions.
It is also possible to obtain the evidence simultaneously. For example, the fingerprint and finger vein information of an individual may be obtained nearly simultaneously by utilizing two scanners housed in the same unit. Multi-biometric Processing sequence – This refers to the order in which the captured information (Multi-biometric acquisition sequence) is processed in order to arrive at a decision. That is, information may be acquired sequentially but processed in parallel or simultaneously. In the serial mode, the processing of information takes place sequentially.
Multi-biometric processing sequences can be divided into two groups, serial and parallel mode. Serial processing requires that multi-biometric information be processed in a sequential order. For example, a multi-biometric system which employees’ fingerprint and face biometrics can operate as follows:
– Process the fingerprint information of the user.
– If the fingerprint sub-system is unable to determine the identity, then the data corresponding to the face biometric is processed.
The arrangement above can significantly reduce the processing time if a decision is made before going through all the biometric subsystems.
Multi-biometric system operating in a serial mode can improve the user’s convenience as well as allow fast and efficient searches in large-scale identification tasks. In the parallel mode, on the other hand, each sub-system processes its information independently at the same time and the processed information is combined using an appropriate fusion scheme. Multi-biometric systems operating in a parallel mode generally have higher recognition accuracy because the system utilizes more evidence about the user for recognition. Multi-biometric systems operating in serial mode can achieve high accuracy as information from multiple sources is progressively accumulated.
It is important for both governments and commercial organisations to have a good understanding of the acquisition sequence and processing sequence in order to ensure that the multi-biometric system performs to its full potential.
Generally, biometric systems are underutilised by governments and commercial organisations due to lack of understanding about the systems.
* Dr Risco Mutelo studied Biometric Engineering at the New Castle University, United Kingdom. He can be contacted at mutelo@namibiabiometricsystems.com