In-Sensor Polarization Convolution Based on Ferroelectric-Reconfigurable Polarization-Sensitive Photodiodes

Area of Science:

• Materials Science
• Optoelectronics
• Nanotechnology

Background:

• In-sensor computing enhances imaging systems by integrating computation within the sensor.
• Existing research has not explored in-sensor light polarization computing strategies.
• Ferroelectric materials offer unique properties for advanced electronic devices.

Purpose of the Study:

• To develop and demonstrate ferroelectric-reconfigurable polarization-sensitive photodiodes (FPPDs) for in-sensor polarization computing.
• To leverage the anisotropic photoresponse of nanowires and ferroelectric reconfigurability for polarization information processing.
• To evaluate the performance of FPPDs in terms of anisotropic ratio and object recognition accuracy.

Main Methods:

• Fabrication of FPPDs using BiFeO3 nanowire arrays.
• Characterization of the devices' anisotropic photoresponse and reconfigurability.
• Implementation of convolution operations using tunable photoresponse for feature extraction.
• Testing object recognition accuracy on road-scene datasets under adverse weather.

Main Results:

• FPPDs demonstrate programmable anisotropic ratios up to 5219, exceeding current technologies.
• The devices successfully perform in-sensor convolution operations on polarization information.
• Object recognition accuracy for road-scene objects under adverse weather conditions is improved to 89.6%.

Conclusions:

• FPPDs represent a significant advancement in in-sensor computing for polarization imaging.
• The developed technology offers a highly efficient vision sensor with potential for intelligent imaging systems.
• This work opens new avenues for exploring ferroelectric materials in advanced optical sensing applications.