Polarization-Aware Deep Object Tracking Polarizasyon Duyarli Derin Nesne Takibi
34th Signal Processing and Communications Applications Conference, SIU 2026, İstanbul, Turkey, 7 - 10 July 2026, (Full Text)
- Publication Type: Conference Paper / Full Text
- Doi Number: 10.1109/siu71813.2026.11637057
- City: İstanbul
- Country: Turkey
- Keywords: AoLP, DoFP camera, DoLP, object tracking, Polarimetric imaging, SiamRPN++
- Hacettepe University Affiliated: Yes
Abstract
Polarization imaging provides additional physical information about scene geometry and material properties that cannot be obtained from conventional intensity-based images. In this study, the JHAD (Joint Histogram of AoLP and DoLP) feature descriptor, which models the joint distribution of polarization magnitude and orientation, is proposed for object tracking in Division-of-Focal-Plane (DoFP) polarization imagery. By jointly modeling the relationship between the Degree of Linear Polarization (DoLP) and the Angle of Linear Polarization (AoLP), polarization characteristics in complex scenes can be represented in a more discriminative manner. The proposed method is evaluated on a DoFP polarization tracking dataset consisting of 105 video sequences captured under various environmental and occlusion conditions. Experimental results demonstrate improved tracking performance, particularly in scenes with occlusion and complex backgrounds.