Low Threshold Voltage Bio-Inspired Visual System Based on InAlZnO: Enabling Multi-Level Information Storage and Motion Trajectory Recognition

In this letter, we report a biomimetic visual perception system based on a 16 × 16 InAlZnO neuromorphic array. Through the directional regulation of oxygen vacancies and the formation of Ag⁺ conductive filaments, the array achieves 64 distinct electrical storage states. It also features a broad spectral response (covering ultraviolet, blue, green, and red light) and a low threshold voltage of ± 0.2 V. Leveraging the multilevel storage capability of the array, we constructed a digital sequence recognition system that attains a recognition accuracy of 95.29% for 10 types of digital sequences. By utilizing the photo responsive forgetting behavior of the array, a system for recognizing object motion trajectories was developed, achieving a recognition rate of 96.30% for 64 different object motion states.