In-sensor image memorization, low-level processing, and high-level computing by using above-bandgap photovoltages
Kun Liu, Shan Tan, Zhen Fan, Haipeng Lin, Jiali Ou, Haoyue Deng, Jinghao Chen, Wenjie Li, Wenjie Hu, Boyuan Cui, Zhiwei Chen, Ruiqiang Tao, Guo Tian, Xubing Lu, Guofu Zhou, Xingsen Gao, Jun‐Ming Liu
Abstract
In-sensor computing holds great promise for ultrafast and energy-efficient machine vision. However, the development of a versatile in-sensor computing system that can integrate image memorization, low-level processing, and high-level computing functions remains a challenge, primarily due to the scarcity of photosensors that can offer both dynamic photoresponse and programmable photoresponsivity. Here, we successfully integrate these multi-functions into a ferroelectric photosensor-based array. The key enabler is the ferroelectric photosensor operating via the bulk photovoltaic effect, which exhibits above-bandgap, dynamically responding, and electrically switchable photovoltages. By using the dynamic photovoltage response, the array is capable of memorizing and pre-processing images, with the ability to adjust the memory and pre-processing effects by ferroelectric polarization. On the other hand, the electrically switchable photovoltages, featuring multi-level switchability and retrievability, enable the array to perform in-sensor high-level computing, achieving 100% accuracy in a 4-class image recognition task (noise level ≤ 10%). Notably, the high precision and reliability of photovoltage-based image memorization and processing greatly benefit from the high photovoltage produced by the ferroelectric photosensor - a distinct advantage for this application. This study lays the foundation for developing versatile in-sensor computing systems that could be utilized across a wide range of machine vision scenarios.