Biometric Security: A Novel Ear Recognition Approach Using a 3D Morphable Ear Model.

Area of Science:

• Biometrics and Cybersecurity
• Computer Vision
• Pattern Recognition

Background:

• Biometrics is crucial for cybersecurity, using physical or behavioral traits for identification.
• Conventional biometric methods, like face or fingerprint recognition, often involve time-consuming registration processes.
• Existing methods for ear-based biometrics can be computationally intensive, limiting real-time applications.

Purpose of the Study:

• To propose a novel and efficient human identification approach using 3D ear images.
• To reduce the computational burden associated with traditional biometric recognition pipelines.
• To enhance the feasibility of biometric authentication for real-time systems.

Main Methods:

• A deep learning algorithm was employed for accurate ear detection in 3D side-face images.
• A statistical 3D morphable ear model (3DMEM) was developed for feature extraction from detected ear images.
• A new recognition algorithm, You Morph Once (YMO), was introduced to minimize probe-gallery registration by comparing stored parameters.

Main Results:

• The proposed method successfully detects ears in 3D images using deep learning.
• The 3DMEM effectively extracts discriminative features from ear images.
• The YMO algorithm significantly reduces computational time by eliminating one-to-one registration, demonstrating practical efficiency.

Conclusions:

• The novel 3D ear-based biometric identification system offers a significant improvement in recognition speed.
• The integration of deep learning, 3DMEM, and the YMO algorithm provides a computationally efficient solution for human identification.
• The proposed method shows promise for real-time biometric applications in cybersecurity.