Hierarchical Blur Identification from Severely Out-of-Focus Images

Abstract

This paper proposes a blur identification method from severely out-of-focus images. The proposed blur identification algorithm can be used in digital auto-focusing and image restoration. Since it is not easy to estimate a point spread function (PSF) from severely out-of-focus images, a hierarchical approach is applied in the proposed algorithm. For severe out of focus blur, the proposed algorithm uses an hierarchical approach for estimating and selecting feasible PSF from successive down sampled images. The down sampled images contain more useful edge information for PSF estimation. The feasible PSF selected, can then be reconstructed for original image resolution level by up sampling methods. In order to reconstruct the PSF accurately, a regularized PSF reconstruction algorithm is used. Finally, we can restore the severely blurred image with the reconstructed PSF. Experimental results show that reconstructed PSF by the proposed hierarchical algorithm can efficiently restore severely out-of-focus images.