Face liveness detection using a light field camera.

Area of Science:

• Computer Vision
• Biometrics
• Image Processing

Background:

• Light field cameras capture ray direction and color, enabling advanced applications beyond conventional imaging.
• Face recognition systems are vulnerable to spoofing attacks using 2D photos and high-definition displays.
• Existing methods struggle to reliably differentiate between genuine and spoofed facial presentations.

Purpose of the Study:

• To propose a novel anti-spoofing method for face recognition systems using light field cameras.
• To leverage unique features extractable from light field images for enhanced security.
• To evaluate the effectiveness of the proposed method against various spoofing techniques.

Main Methods:

• Utilizing a light field camera to capture facial data, recording both color and directional information of light rays.
• Extracting two distinct features from raw light field photographs by analyzing them from different viewpoints.
• Developing and employing light field photograph databases for rigorous experimental validation.

Main Results:

• The proposed method demonstrates high accuracy in detecting spoofing attacks, ranging from 94.78% to 99.36%.
• The extracted features proved effective in distinguishing between genuine facial presentations and spoofing attempts like 2D photos and tablet images.
• Experimental results confirm the robustness of the approach across different types of spoofing attacks.

Conclusions:

• Light field cameras offer a promising solution for robust face anti-spoofing in biometric systems.
• The novel feature extraction technique significantly enhances the security of face recognition against presentation attacks.
• The developed method provides a reliable and accurate defense against common spoofing tactics.