Enhanced facial recognition for thermal imagery using polarimetric imaging

Area of Science:

• Optics and Photonics
• Biomedical Imaging
• Computer Vision

Background:

• Conventional thermal imaging (MidIR or LWIR) struggles with human identification due to the "ghosting" effect and limited spatial detail.
• Polarization-state information in thermal imaging has not been fully utilized for facial feature enhancement.
• Extracting subtle surface features is crucial for reliable identification in thermal imagery.

Purpose of the Study:

• To investigate the utility of long-wave-infrared (LWIR) polarimetric imaging for enhanced facial feature representation.
• To demonstrate how polarization-state information can improve human identification from thermal images.
• To present a novel approach for overcoming limitations of conventional thermal imaging in facial recognition.

Main Methods:

• Acquisition of long-wave-infrared (LWIR) polarimetric-based thermal images of facial profiles.
• Retention and display of polarization-state information, including Stokes images (S0, S1, S2) and degree-of-linear-polarization (DoLP) images.
• Comparison of polarimetric thermal images with conventional thermal imagery to assess feature enhancement.

Main Results:

• Polarimetric thermal images reveal enhanced facial features, textures, and details not visible in conventional thermal imagery.
• The "ghosting" effect, a common issue in conventional thermal imaging, is mitigated by using polarimetric information.
• Subtle surface features of the human face are effectively extracted using polarimetric data, leading to improved identification.

Conclusions:

• LWIR polarimetric imaging offers a significant advantage over conventional thermal imaging for facial feature analysis and identification.
• The incorporation of polarization-state information is key to overcoming the limitations of traditional thermal imaging for biometric applications.
• This technique shows promise for enhancing human identification capabilities in scenarios where conventional thermal imaging is insufficient.