Single-pixel imaging through random media with automated adaptive corrections

The existence of random media is challenging in optical imaging, as the existing approaches usually cannot work well when the optical channel exhibits a certain level of randomness. Here, we report an automated adaptive correction scheme for single-pixel imaging through random media. An alternating projection method is developed to reconstruct an object from light intensities recorded by a single-pixel detector. A series of scaling factors are incorporated into object reconstruction to correct wave distortions induced by random media. With the introduced scaling factors, an essential relationship between collected and theoretical light intensities is revealed. It is illustrated that the proposed corrections on the realizations do not require prior knowledge about random media, and can be adapted to various real-world scenarios. High-quality imaging through random media can always be realized in experiments, and the proposed approach opens up an avenue for high-quality imaging through random media in various applications.