Novel Approaches to Improve Iris Recognition System Performance Based on Local Quality Evaluation and Feature Fusion

Table 9

Description of process fragile bits.

Methodology	Process method of fragile bits	Results
Hollingsworth et al. [5, 36]	Masking iris code bits corresponding to complex filter response near the axes of the complex plane	Improve the separation between the match and nonmatch Hamming distance distributions
Hollingsworth et al. [8]	Consider the fragile bits provide beneficial information rather than ignoring fragile bits completely	Score fusion of fragile bit distance and Hamming distance works better than Hamming alone, and it improved the accuracy of matches
Hollingsworth et al. [37]	Masks fragile bits	𝑑^′ value is 8.2516 compared with 7.4825 for nonmask fragile bits
Bolle et al. [7]	Theoretically proved the existence of fragile bits	—
Dozier et al. [38]	Used genetic search to minimize the number of iris code bits	Approximately reduce 89% of iris code bits and discard the fragile bits
Dozier et al. [39]	Only kept bits that were 90% or 100% consistent	Discard the fragile bits
Proposed	Divided normalized iris into multitracks, and fused all tracks by local quality evaluation with PSO	𝑑^′ value increase and EER decrease