Active headrest combined with a depth camera-based ear-positioning system

Area of Science:

• Acoustics
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Active headrests (AHR) provide localized noise reduction (NR) but are sensitive to head position.
• Existing head tracking systems primarily address translational head movements, limiting robustness against combined motions.
• Maintaining effective NR requires precise alignment between the AHR quiet zone and the listener's ear.

Purpose of the Study:

• To develop and evaluate an ear-positioning (EP) system integrated with AHR for robust noise reduction.
• To enhance the performance of AHR against both translational and rotational head movements.
• To improve the overall effectiveness of AHR in real-world scenarios with dynamic head motion.

Main Methods:

• An ear-positioning (EP) system was developed using a depth camera and a human pose estimation model.
• The EP system was integrated with active headrests (AHR).
• A post-processing algorithm was implemented to handle occluded ear scenarios.

Main Results:

• The EP system demonstrated effective tracking of ear movements during translation and rotation.
• AHR integrated with the EP system achieved substantial noise reduction (up to 11.7/12.2 dBA for translation, 11.4/13.6 dBA for rotation).
• Performance improvements were significant compared to AHR without the EP system (e.g., -5.7/-6.9 dBA without system).

Conclusions:

• The proposed ear-positioning system significantly enhances the robustness and performance of active headrests.
• The integration of EP with AHR provides effective noise reduction across a range of head movements, including translation and rotation.
• This technology offers a promising solution for maintaining personalized quiet zones in dynamic environments.