PSF Estimation of Simple Lens Based on Circular Partition Strategy.

Recently, single-lens computational imaging has gradually become a new research direction of computational photography, which combines the front-end simple optical imaging equipment with the late image restoration algorithm to obtain high-quality images. Single lens computational imaging is essentially a problem of image restoration. The estimation accuracy of point spread function (PSF) will directly affect the quality of image restoration. Existing spatially variant PSF estimation methods usually divide images into rectangular blocks. Considering the imaging characteristics of single lens, a PSF estimation method based on a circular partition strategy is proposed in this paper. Experimental results show that this segmented method can achieve better PSF estimation accuracy and improve image restoration quality.