SELF-SUPERVISED ERROR REDUCTION OF FAST RESTORATION OF UNKNOWN BLURRED NOISY IMAGES

When images are transmitted through optical systems, they usually suffer from a combination of blurring and noise. When blurring is unknown and the noise varies over th, neighbourhood, image restoration is more difficult. In light of this problem, an approach for the restoration of unknown blurred images in the presence of noise was developed. In the Letter, the effectiveness and practicality of this technique is extended using a self-supervised rule.