Abstract
The increasing demand for visual data processing reveals the limitations of traditional electronic systems in speed, energy efficiency, and adaptability. While optical computing offers a promising alternative, current systems often lack the flexibility required for multispectral and adaptive visual tasks. Here, we present an efficient way for highly adaptive, multispectral image filtering based on active Tamm plasmon resonators. We realize precise control over resonant wavelengths, enabling selective spectral targeting with high-quality factors. To achieve a gradual on/off function in the Tamm plasmon resonator, we integrate PEDOT:PSS, whose doping state modulates its metallic or dielectric properties. The partial doping of PEDOT:PSS allows for memorizing states, facilitating long-term potentiation and depression, and is essential for forming multiple synaptic states. By combining the high modulation depth, theoretically reaching 99%, with the non-volatile nature of PEDOT:PSS, we achieve stable multiple synaptic states with subtle saturation, resulting in 256 stable synaptic weights.