An optical meta-image-processor for enhanced imaging through strongly scattering media.

Strongly scattering conditions are detrimental to imaging. Conventional methods rely solely on post-processing techniques to enhance the ballistic components of an obscured target to recover the image; however, they struggle in complex conditions where the ballistic components are heavily overwhelmed by scattering. In this work, we present an optical meta-image-processor (MIP) that tailors the scattered point spread function of the imaging system to enable high-quality, deep imaging through strongly scattering media. The MIP performs both Laplacian and Gaussian operations in a single device, effectively suppressing background interference and Gaussian noise in the obscured image. Experimental results demonstrate that clear information can be recognized with the MIP, even when the optical thickness of the scattering medium reaches a challenging value of 17.05. Without the MIP, such imaging depth cannot be achieved through direct imaging, even when combined with any other post-processing techniques. Additionally, the MIP shows potential for enhancing the diagnostic performance of fundus cameras in the presence of cataracts. Both simulated and experimental evidence confirms the MIP's capability to reveal hidden information under strongly scattering conditions, underscoring its promising potential to enhance imaging depth in biomedical imaging, machine vision, and artificial intelligence, especially in complex environments.