# Hardware Implementation of Multimodal Biometric using Fingerprint and Iris

**Authors:** Tariq M Khan  
**arXiv:** [2201.05996](https://arxiv.org/abs/2201.05996)  
**Date:** 2022-01-16  
**Summary page:** https://www.summarizepaper.com/en/paper/2201.05996v1/

## Key points

- Tariq M Khan presents a hardware implementation of a multimodal biometric system combining fingerprint and iris traits.
- The system optimizes each biometric trait at the software level to address False Acceptance Rate (FAR) and False Rejection Rate (FRR) issues.
- Promising results are achieved with a FAR of 1.97% and FRR of 12.79%.
- Iris recognition is highlighted as more accurate than fingerprint recognition in literature, leading to a weightage of 0.4 for fingerprint and 0.6 for iris in the fusion process.
- Hardware implementation uses a low-cost Cyclone IV GX P4CGX150F31 FPGA for efficient software-hardware co-designs.
- The proposed iris recognition system consumes approximately 12K logic elements, accounting for only 12% of total logic elements available, and about 10% logic registers.
- Fingerprint recognition consumes more logic cells and registers but less memory bits than iris recognition due to differences in image processing requirements.
- The study contributes to advancing multimodal biometric systems by presenting a novel hardware architecture integrating fingerprint and iris traits for enhanced security applications.

## Summary

Tariq M Khan presents a hardware implementation of a multimodal biometric system that leverages the parallelism inherent in the system. The proposed system combines two biometric traits, fingerprint and iris, using a classical fusion technique at the matching-score level. By optimizing each biometric trait at the software level to address issues affecting False Acceptance Rate (FAR) and False Rejection Rate (FRR), the author achieves promising results with a FAR of 1.97% and FRR of 12.79%. The study highlights the importance of iris recognition over fingerprint recognition due to higher accuracy rates in literature. To reflect this, a weightage of 0.4 is assigned to fingerprint and 0.6 to iris in the fusion process. The hardware implementation utilizes a low-cost Cyclone IV GX P4CGX150F31 FPGA, combining an Intel embedded processor with an Altera Cyclone IV GX FPGA for efficient software-hardware co-designs. Detailed analysis of hardware resource utilization reveals that the proposed iris recognition system consumes approximately 12K logic elements, accounting for only 12% of total logic elements available, and about 10% logic registers. In comparison, the fingerprint recognition system consumes more logic cells and registers but less memory bits than the iris recognition due to differences in image processing requirements. Overall, this study contributes to advancing multimodal biometric systems by presenting a novel hardware architecture that effectively integrates fingerprint and iris traits for enhanced security applications.