Computational workload in biometric identification systems: an overview

Abstract

Computational workload is one of the key challenges in biometric identification systems. The naïve retrieval method based on an exhaustive search becomes impractical with the growth of the number of the enrolled data subjects. Consequently, in recent years, many methods with the aim of reducing or optimising the computational workload, and thereby speeding-up the identification transactions, in biometric identification systems have been developed. In this article, a taxonomy for conceptual categorisation of such methods is presented, followed by a comprehensive survey of the relevant academic publications, including computational workload reduction and software/hardware-based acceleration. Lastly, the pertinent technical considerations and trade-offs of the surveyed methods are discussed, along with an industry perspective, and open issues/challenges in the field.

Multi-Biometric Identification With Cascading Database Filtering

“…Since the overall computational costs in a biometric identification scenario are dominated by performing the biometric comparisons (see e.g. Drozdowski et al [24]), most computational workload reduction approaches are aimed at that step in the system pipeline. It should be noted, that due to certain properties of biometric data (i.e.…”

“…An exhaustive survey of this research area is out of scope for this article -for more details, the interested reader is referred to other works on this topic. Specifically, in a recently published work of Drozdowski et al [24], a biometric characteristic-agnostic, concept-based taxonomy of computational workload reduction approaches in biometrics has been proposed. Additionally, the authors conducted a comprehensive survey of computational workload reduction in biometric identification systems in the context of said taxonomy.…”

“…In addition to the aforementioned biometric performance improvement, the proposed system has an impact on the computational complexity of a biometric identification transaction. [24] for a more detailed discussion on this topic). The main reason for this are the different feature representations and comparators across the modalities.…”

“…In recent years, a significant research effort has been devoted to addressing this topic by developing methods for computational workload reduction in biometric systems (see subsection 2.2 and a recent survey of Drozdowski et al [24] for more details). The contribution of this work in this context is a proposal of an information fusion scheme, as well as an experimental evaluation thereof on a large compound dataset in the biometric open-set identification scenario.…”

A multimodal retina‐iris biometric system using the Levenshtein distance for spatial feature comparison

“…The reliability tests carried out on the unimodal recognition systems described so far offer good performance in terms of safety and robustness, although the two systems are not yet adequate to be considered truly reliable (Tables 7 and 8). In order to select the best decision threshold S 1 , the FAR and FRR values were calculated by varying S 1 in the range [20,70], with a step of 5.…”