.png){kind=icon}
.png){kind=icon}
Return
Retina-Inspired Flexible Visual Synaptic Device for Dynamic Image Processing
.png){kind=icon}
PRE
AI
DOI：10.1021/acsami.4c16378.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Exploiting biomimetic perception of invisible spectra in flexible artificial human vision systems (HVSs) is crucial for real-time dynamic information processing. Nevertheless, the fast processing of motion objects in natural environments poses a challenge, necessitating that these artificial HVSs simultaneously have swift photoresponse and nonvolatile memory. Here, inspired by the human retina, we propose a flexible UV neuromorphic visual synaptic device (NeuVSD) based on GaO x @GaN-composited nanowires for dynamic visual perception. Benefiting from the combined action of oxygen adsorption kinetics on the GaO x shell and photoelectronic excitation in the GaN core, the NeuVSD can achieve rapid photoresponse and long-term plasticity simultaneously, a feature that surpasses traditional devices based on persistent photoconductivity mechanisms. Beyond the optoelectronic synaptic plasticity, the flexible NeuVSD on the PET/PI substrate exhibits good performance stability after 1000 bending cycles, profiting from the high adaptability of nanowires. Furthermore, target recognition and motion detection with weak light intensities are achieved through the establishment of neuromorphic visual neural networks, relying on dynamic exposures and edge information processing, respectively. Real-time edge detection can still be realized under a 40% noise factor and effectively remove the noise background. The flexible NeuVSD array-based artificial visual system can enhance the capability of dynamic visual information processing in low-light and high-noise conditions, thereby fostering the evolution of in-sensor computing and artificial intelligence.
Keywords:
nanowire
synaptic plasticity
neuromorphicvisual synaptic device
optoelectronic synapse
human visual system
Journal
IF:
8.2
Papers:
6.1W
Citations:
38.7W