神经形态工程学
光电子学
计算机科学
二极管
量子点
紫外线
电致发光
材料科学
生物光子学
能量(信号处理)
人工神经网络
发光二极管
图层(电子)
物理
RGB颜色模型
光学
噪音(视频)
人工智能
弹道
图像处理
显示设备
光电二极管
随机存取
机器视觉
出租车
激光器
计算机视觉
运动检测
彩色视觉
闪烁
激光二极管
电子工程
作者
Haohong Jiang,Lixiang Chen,Yuanhong Hu,Tongzhou Wei,Qiaoming Zhang,Yuantao Zeng,Shu‐Jen Wang,Furong Zhu,Feng Wang,Yanlian Lei
摘要
Artificial vision systems conventionally segregate sensing, memory, processing, and display into discrete modules, incurring latency, energy penalties, and architectural complexity. Here, we report a miniaturized synaptic quantum-dot light-emitting diode that monolithically integrates optical sensing, charge-storage memory, and visible emission. Featuring a dual-spectrum-sensitive hole transport layer responsive to ultraviolet (UV) and near-infrared (NIR) light, the device achieves wavelength-gated synaptic plasticity. UV stimuli generate minute-scale electroluminescent afterimages via deep-level hole trapping, while NIR pulses enable millisecond-scale relaxation. This spectral asymmetry establishes distinct “write” and “develop/erase” pathways, facilitating privacy-preserving imaging without external circuitry. Furthermore, the device exhibits real-time motion trajectory display with 98.8% direction recognition accuracy via neural network validation and hardware-level adaptive noise filtering that suppresses random spikes while preserving correlated signals. By transforming the emissive layer into an on-device processing element, this architecture removes the need for separate sensor, processor, and display components, offering a compact platform for intelligent vision systems operating beyond the visible spectrum.
科研通智能强力驱动
Strongly Powered by AbleSci AI