A flexible material for low-light artificial visual perception
DOI: 10.1063/10.0044497
A flexible material for low-light artificial visual perception lead image
Memristors are a key component for developing flexible neuromorphic visual systems. Their ability to quickly tune their resistance based on optoelectronic inputs mimics biological signal weight adjustment, essential for artificial retina-like visual sensors.
By creating a composite material of perovskite quantum dots and graphene oxide, Yue et al. demonstrated a memristor with excellent promise as a next-generation flexible neuromorphic system.
Their material behaves differently in dark and lit conditions. Under dim illumination, the memristor’s conductance range expands, allowing it to better filter noise and, in effect, amplify faint signals. By processing these low-contrast images, the material outputs clear silhouettes and allows features to be extracted for low-light object recognition.
“This device-level tuning translates directly to advanced neuromorphic vision performance,” said author Jingjuan Wang. “For noisy, low-light images, recognition accuracy rises over 10%, signal-to-noise ratio doubles, and foreground/background signals are easily separated — all processed locally on the flexible sensor without extra circuits.”
Moreover, the authors’ quantum dot/graphene oxide memristor is flexible. In tests, it continued to work reliably after thousands of bends. Because it resists cracking under deformation, it lends itself well to wearable applications; it can be attached to curved skin to capture weak, ambient light signals.
To extend the work into real-time image classification, the researchers are planning to combine their flexible memristor arrays with low-power circuits and add polarization and wavelength-specific responses to mimic human color vision.
Source: “Flexible optoelectronic memristor with photo-enhanced resistive switching for low-light visual perception,” by Ziwei Yue, Siyu Zhao, Yuchun Li, Lingzhi Tang, Shuxia Ren, and Jingjuan Wang, Applied Physics Letters (2026). The article can be accessed at https://doi.org/10.1063/5.0339447 .
