Biomimetic visual perception system based on InAlZnO optoelectronic neuromorphic array for static image processing and dynamic trajectory perception

Vision is a vital means for humans to perceive the environment, with 80% of information from visual perception. Developing visual systems approaching or surpassing human-level vision plays an indispensable role in advancing autonomous driving, intelligent security, and other fields. Optoelectronic neuromorphic devices, integrating sensing and memory, are promising for emulating human vision. This study constructs a biomimetic visual system based on a 16 × 16 InAlZnO optoelectronic neuromorphic array with oxygen vacancy gradients. Under multi-factor modulation (oxygen vacancy concentration differences, Ag ions, heterojunction interfaces), the device achieves electrical/optical conductance tunability. Integrated with external circuits and a field-programmable gate array, the system successfully emulates human vision capabilities: image memory, denoising, attention mechanism, and motion state perception (direction, speed, color). Image digit recognition based on visual attention reaches 97.15% accuracy, and motion state recognition reaches 100%. This system will promote bionic vision development and application, paving the way for high-performance neuromorphic vision systems surpassing the human eye.