Detail Recovery for Single-image Defocus Blur

Abstract

We presented an invited talk at the MIRU-IUW workshop on correcting photometric distortions in photographs. In this paper, we describe our work on addressing one form of this distortion, namely defocus blur. Defocus blur can lead to the loss of fine-scale scene detail, and we address the problem of recovering it. Our approach targets a single-image solution that capitalizes on redundant scene information by restoring image patches that have greater defocus blur using similar, more focused patches as exemplars. The major challenge in this approach is to produce a spatially coherent and natural result given the rather limited exemplar data present in a single image. To address this problem, we introduce a novel correction algorithm that maximizes the use of available image information and employs additional prior constraints. Unique to our approach is an exemplar-based deblurring strategy that simultaneously considers candidate patches from both sharper image regions as well as deconvolved patches from blurred regions. This not only allows more of the image to contribute to the recovery process but inherently combines synthesis and deconvolution into a single procedure. In addition, we use a top-down strategy where the pool of in-focus exemplars is progressively expanded as increasing levels of defocus are corrected. After detail recovery, regularization based on sparsity and contour continuity constraints is applied to produce a more plausible and natural result. Our method compares favorably to related techniques such as defocus inpainting and deconvolution with constraints from natural image statistics alone.

A regularization LR-algorithm for restoring images on Gaussian noises model

“…The sub-window size is 4x4. The residual signal related 423 parameter a is set as 0.2�0.5 and y is set to 0.1 �0.4 in formula (10). Iterative number is 120.…”

“…Therefore, we have the gauss model LR algorithm equation using residual signals regularization constraints: fk +l = fk +h* [ a<P)loe +(l-a)<pjUoj +r<ptUot JR k (10) Since the value of marginal area corresponding to <P e *coe is great, it can highlight the edge and reduce the smoothness. On the other hand, <PI *0)1 corresponding to flat area enhance the smoothing effect and suppress the ringing effect.…”

Sharpening Out of Focus Images using High‐Frequency Transfer

“…Deconvolution methods addressing out‐of‐focus blurs typically introduce ringing artifacts [Ric72]. [TTBL09] present a natural image statistics prior for deconvolution to enhance details. Because of the limited image statistical priors, the method does not work for general natural images, as stated by the authors.…”

Automatic foreground-background refocusing

“…Their method requires enough sharp step edges in the images. [6] and [7] exploit statistical similarities between the blurry and sharp parts of an image to guide the deconvolution process to refocus images.…”