Memristive retinal biomimetic devices and their light adaptation characteristics

The human visual system, relying on the synergistic action of cone and rod cells, achieves remarkable light adaptation across a light intensity range spanning approximately 10 orders of magnitude (about 120 dB). Its core mechanisms include the slow regeneration of rhodopsin during dark adaptation and the rapid response dominated by cone cells during light adaptation. Inspired by this biological characteristic, this study proposes and develops a memristive retinal biomimetic photomodulator (MRPM). Constructed based on an ITO/PEDOT/P3HT:PCBM/SnO2/Ag layered structure, this device exhibits biomimetic dynamic light response characteristics: under low light conditions, the photocurrent gradually increases over time; under high light conditions, the photocurrent rapidly decays and tends to stabilize. The device covers a broad spectral response range of 300–850 nm, exceeding the visible light band of the human eye (380–780 nm). Experiments show that the MRPM achieves adaptive image enhancement and contrast adjustment within a 120 dB dynamic range, effectively simulating the dark/light adaptation process of the retina. Furthermore, this study constructs a hybrid neuromorphic vision architecture that combines MRPM with convolutional neural networks (CNNs), significantly improving the robustness and recognition accuracy of machine vision systems under extreme lighting conditions. This work provides a new research paradigm for the design and implementation of biomimetic vision sensors and adaptive artificial retina systems.