AI technology for e-paper displays

Electrophoretic Electronic Paper (EPD), as a typical reflective display technology, is widely used in various low-power display scenarios. To achieve multiple gray scales in color electrophoretic displays typically requires halftoning-based image processing. While this approach enables visual effects consistent with the original image using limited specific colors, it leads to a significant increase in driving current, which runs counter to the low-power consumption characteristic of electronic paper. Furthermore, during dynamic EPD operation, the hysteresis behavior of electrophoretic particles gives rise to pronounced ghosting artifacts. Traditional solutions mainly rely on repeated manual tuning of driving parameters or adopting fixed control strategies based on past experience. However, these methods are not only time-consuming and labor-intensive but also difficult to adapt to variable environmental conditions, and even less able to achieve a balance between conflicting optimization objectives. Addressing the above challenges, this paper separately discusses application schemes of artificial intelligence technology on these two typical problems. In terms of power optimization, this paper utilizes Generative Adversarial Networks (GAN) combined with multi-objective optimization algorithms to generate lowpower halftone image patterns; in terms of ghosting suppression, an automatic ghosting recognition and closed-loop waveform design method based on Convolutional Neural Networks (CNN) is proposed. Experimental results show that the former reduces driving power consumption by 43% under the premise of ensuring visual quality, and the latter achieves a 92% ghosting suppression rate, proving that AI technology has significant application value in improving the energy efficiency and display quality of electronic paper.