13 - Multibiometrics

Multibiometric Systems

Multibiometric systems combine information from multiple biometric traits (such as fingerprints, facial features, iris patterns) to enhance accuracy and reliability. This integration mitigates the limitations of individual biometric systems and improves overall performance.

Main Techniques in Multibiometric Systems

  • Feature-Level Fusion: Combines raw biometric data from multiple sources before matching. For instance, features from fingerprints and facial images are merged.
  • Score-Level Fusion: Integrates matching scores obtained from individual biometric systems to make a final decision. It can use techniques like averaging, weighted sum, or decision-level fusion methods.

Score-Level Fusion

  • Averaging: Simplest form of score fusion where scores from different biometric systems are averaged to obtain a final score. It balances out variations and errors in individual scores.
  • Weighted Sum: Assigns specific weights to scores based on the reliability or performance of individual biometric systems. Weighted sum offers flexibility by emphasizing the more accurate systems.
  • Decision-Level Fusion: Makes the final decision based on the decisions made by individual biometric systems. It can be performed using techniques like majority voting, weighted voting, or logical rules.

How Score-Level Fusion Works

  • Score Collection: Matching scores are collected from individual biometric systems after analyzing the input data (e.g., fingerprints, facial features).
  • Normalization: Scores may be normalized to ensure consistency and comparability, especially if different biometric modalities are used.
  • Fusion Algorithm: The selected fusion algorithm (averaging, weighted sum, or decision-level rules) combines the normalized scores.
  • Thresholding: The fused score is compared against a predefined threshold. If it exceeds the threshold, the system authenticates the user.

Benefits of Multibiometric Systems with Score-Level Fusion

  • Improved Accuracy: Combining scores from multiple sources reduces errors and enhances overall system accuracy.
  • Enhanced Security: Multibiometric systems offer higher security levels due to the complexity involved in spoofing or falsifying multiple biometric traits simultaneously.
  • Increased Robustness: Multibiometric systems are more robust against environmental factors (e.g., lighting conditions) as different biometric traits respond differently to such factors.