Long-Term Dynamic Characteristics of CsPbBr3 Quantum Dot Memristors for Offline Image Enhancement in Brain–Computer Interfaces

Long-term deployment of memristor-based image enhancement in wearable or implantable brain–computer interfaces (BCIs) remains challenging due to the stringent requirements for device reliability and stability. Here, we leverage the long-term dynamic physical characteristics of memristors to achieve offline image enhancement for BCI applications. Highly crystalline and uniform-sized CsPbBr3 quantum dots are synthesized via a modified thermal injection method and integrated into an ITO/PQDs/Ag memristor. The device exhibits a voltage-dependent resistive switching: a pristine low-resistance state at 0.5 V without hysteresis transitions to stable bipolar switching at 1 V, attributed to the in-situ formation of an interfacial AgBrx layer, with the switching governed by the synergistic interplay of Ag+ filament dynamics and Br– interfacial accumulation. From the device stability and cycling data, two physical parameters, namely, low-resistance state stability and transition agility, are extracted and mapped to sharpening and contrast intensity factors in image enhancement algorithms, respectively. This method effectively improves the detail clarity and contrast of blurred images, as evaluated by a convolutional neural network, where recognition accuracy improves by 16.8% compared to original blurred images. This work simplifies the complex parameter tuning associated with traditional enhancement algorithms, offering a pathway for integrating memristor-based image preprocessing into wearable and implantable BCI systems.