Bioinspired Adaptive Neuron Enabled by Self-powered Optoelectronic Memristor and Threshold Switching Memory for Neuromorphic Visual System.

Visual adaptation allows organisms to effectively analyze visual information in varying light conditions by autonomously adjusting photosensitivity, which is essential for the visual system to perform accurate perception in complex environments. In order to realistically implement the functionality of the visual system, the exploration of bioinspired electronics with adaptive capability is highly desired. Herein, a self-powered optoelectronic memristor based on ZnO/WO heterojunction is developed, which can exhibit the visual adaptation functions of desensitization and Weber's law. These functions are achieved through the coupling of the photovoltaic effect with electron trapping in the space charge region of the heterojunction. Furthermore, a bioinspired visual adaptive neuron has been constructed, comprising an optoelectronic memristor and a NbO-based threshold switching memory, capable of directly converting constant light stimuli into dynamic spike trains. Finally, the adaptive image preprocessing is realized, which promotes the improvement of the object recognition accuracy during the overexposed image recognition process. This study offers a novel approach to developing biologically plausible visual adaptation, fostering the future progress of dynamic neuromorphic visual systems.