ScatT-LOOP: scattering tetrolet-LOOP descriptor and optimized NN for iris recognition at-a-distance

Area of Science:

• Computer Science
• Biometrics
• Artificial Intelligence

Background:

• Iris Recognition at a Distance (IAAD) faces significant challenges due to poor image quality in dynamic environments, impacting recognition accuracy.
• Existing methods struggle with the defects inherent in visual imaging for IAAD.
• Effective IAAD is crucial for security and identification applications.

Purpose of the Study:

• To propose a new, effective method for Iris Recognition at a Distance (IAAD).
• To enhance the accuracy and robustness of iris recognition systems in challenging conditions.
• To introduce the Chronological Monarch Butterfly Optimization (Chronological MBO)-enabled Neural Network (NN) for IAAD.

Main Methods:

• Developed a Chronological MBO algorithm by integrating Chronological theory with Monarch Butterfly Optimization (MBO).
• Employed automatic iris image segmentation and normalization using Hough Transform (HT) and Daugman's rubber sheet model.
• Utilized a novel ScatT-LOOP descriptor, integrating scattering transform (ST), Local Optimal Oriented Pattern (LOOP), and Tetrolet transform (TT) for feature extraction.

Main Results:

• The proposed Chronological MBO-enabled NN achieved a high accuracy of 0.97.
• The method demonstrated a low False Acceptance Rate (FAR) of 0.005 and False Rejection Rate (FRR) of 0.005.
• Experimental results on the CASIA Iris dataset confirmed the superiority of the proposed method over existing techniques.

Conclusions:

• The Chronological MBO-enabled NN offers a significant advancement in IAAD.
• The ScatT-LOOP descriptor effectively captures essential texture and orientation details for robust recognition.
• The proposed method provides a more accurate and reliable solution for iris recognition in real-world scenarios.