High-Resolution Metalens Imaging with Sequential Artificial Intelligence Models

An analysis of the optical response of a GaN-based metalens was conducted alongside the utilization of two sequential artificial intelligence (AI) models in addressing the occasional issues of blurriness and color cast in captured images. The optical loss of the metalens in the blue spectral range was found to have resulted in the color cast of images. Autoencoder and CodeFormer sequential models were employed in order to correct the color cast and reconstruct image details, respectively. Said sequential models successfully addressed the color cast and reconstructed details for all of the allocated face image categories. Subsequently, the CIE 1931 chromaticity diagrams and peak signal-to-noise ratio analysis provided numerical evidence of the AI models' effectiveness in image reconstruction. Furthermore, the AI models can still repair the image without blue information. Overall, the integration of metalens and artificial intelligence models marks a breakthrough in enhancing the performance of full-color metalens-based imaging systems.