Key points are not available for this paper at this time.
Abstract Capturing images with high fidelity that can discern objects in scenes across diverse lighting conditions is crucial for machine vision systems. Despite great efforts that have been adopted in conventional silicon‐photodiode‐based image sensors to achieve a high dynamic range (HDR), their dynamic range is still limited. Inspired by the dynamic adaptability of retinal cone/rod cells, a novel dual‐mechanism bionic vision sensor based on silicon‐on‐insulator and graphene stacks is reported. Under reverse bias, the device functions as a cone cell in photodiode mode, while under forward bias, it transitions to photoconductor mode, mimicking the behavior of a rod cell. This robust switch between photodiode and photoconductor modes mimics the complementary versatility of cone and rod cells in diverse lighting scenarios, achieving an expansive dynamic perception range of up to 170 dB. With response times of 5 ns and 4 µs in photodiode and photoconductor modes, respectively, the device fulfills the rapid response requisites of machine vision. Furthermore, imaging fusions are demonstrated by combining mean filtering based on images captured by the detector operating at different modes. This pioneering development offers a new device architecture that elevates the performance of photodetectors in image sensors for future machine vision systems.
Zhou et al. (Thu,) studied this question.