The paper discusses the use of fusion of different comparators to enhance periocular recognition when images from different smartphones are compared. This is due to the proliferation of cameras and personal devices which result in a wide variability of imaging conditions. The authors use a probabilistic fusion framework based on linear logistic regression, obtaining a reduction in cross-sensor EER of up to 40% due to the fusion. The framework also provides a simple solution to handle signals from different devices, eliminating the need for sensor-specific thresholds.
Publication date: 3 Nov 2023
Project Page: Not provided
Paper: https://arxiv.org/pdf/2311.01237
