Image Zoom Completion

Abstract

We consider the problem of recovering a high-resolution image from a pair consisting of a complete low-resolution image and a high-resolution but incomplete one. We refer to this task as the image zoom completion problem. After discussing possible contexts in which this setting may arise, we introduce a nonlocal regularization strategy, giving full details concerning the numerical optimization of the corresponding energy and discussing its benefits and shortcomings. We also derive two total variation-based algorithms and evaluate the performance of the proposed methods on a set of natural and textured images. We compare the results and get with those obtained with two recent state-of-the-art single-image super-resolution algorithms.

Texture Reconstruction Guided by a High-Resolution Patch

“…There exist few methods in the literature that consider the texture synthesis problem guided by a low-resolution texture [46], [47]. These approaches have some similarities with the proposed one with notable important differences:…”

“…• In [46], the problem is solved by taking into account a nonlocal model. The approach consists in connecting each missing pixel to a set of neighboring pixels and some other pixels in the high-resolution patch.…”

“…Regarding the problem considered in this paper (recall Fig. 1), an extension of NLTV regularization was recently proposed in [46]. This prior consists in building a pointwise estimate of the image, where each missing pixel is connected to a local neighborhood and to the most similar pixels in the given high resolution patch.…”

Texture synthesis guided by a low-resolution image

“…Regarding the problem considered in this paper, the authors in [15] propose to recover the high-resolution image from a pair of images: a complete low-resolution image and a highresolution but incomplete one (sample). The resulting problem is solved using an extension of the nonlocal total variation model, where a set of connections is built between the missing high-resolution pixels and a set of pixels that lies in the sample.…”

“…The work in [17] exploits a patch-based nonlocal regularization and a Generalized Gaussian model of the texture gradients, whose parameters are learned on the high-resolution patch. Differently from [15] and [16], the authors in [17] use a prior on the spatial covariance of the synthesized image.…”