Improving the quality of degraded images is a key problem in image processing, but the breadth of the problem leads to domain-specific approaches for tasks such as super-resolution and compression artifact removal. Recent approaches have shown that a general approach is possible by learning application-specific models from examples; however, learning models sophisticated enough to generate high-quality images is computationally expensive, and so specific per-application or per-dataset models are impractical. To solve this problem, we present an efficient semi-local approximation scheme to large-scale Gaussian processes. This allows efficient learning of task-specific image enhancements from example images without reducing quality. As such, our algorithm can be easily customized to specific applications and datasets, and we show the efficiency and effectiveness of our approach across five domains: single-image super-resolution for scene, human face, and text images, and artifact removal in JPEG- and JPEG 2000-encoded images.

In this paper, we describe a framework for learning-based image enhancement. At the core of our algorithm lies a generic regularization framework that comprises a prior on natural images, as well as an application-specific conditional model based on Gaussian processes. In contrast to prior learning-based approaches, our algorithm can instantly learn task-specific degradation models from sample images which enables users to easily adopt the algorithm to a specific problem and data set of interest. This is facilitated by our efficient approximation scheme of large-scale Gaussian processes. We demonstrate the efficiency and effectiveness of our approach by applying it to two example enhancement applications: single-image super-resolution as well as artifact removal in JPEG-encoded images.

This paper proposes a framework for single-image super-resolution. The underlying idea is to learn a map from input low-resolution images to target high-resolution images based on example pairs of input and output images. Kernel ridge regression (KRR) is adopted for this purpose. To reduce the time complexity of training and testing for KRR, a sparse solution is found by combining the ideas of kernel matching pursuit and gradient descent. As a regularized solution, KRR leads to a better generalization than simply storing the examples as has been done in existing example-based algorithms and results in much less noisy images. However, this may introduce blurring and ringing artifacts around major edges as sharp changes are penalized severely. A prior model of a generic image class which takes into account the discontinuity property of images is adopted to resolve this problem. Comparison with existing algorithms shows the effectiveness of the proposed method.

This paper presents a text image super resolution algorithm based on total variation (TV). Text images typically consist of slim strokes on background. Thus, there are three different local characteristics as homogeneous, directed and complex on text image. Homogeneous region corresponds to background and directed means the region with dominant stroke direction and remaining is complex region. We proposed higher order smoothing on homogeneous region and anisotropic regularization on directed region which encodes the preference of edge direction by smoothing along preferred direction only. Required regularization terms are combined in proposed anisotropic TV functional and controlled by relating parameters. We calculated relating parameters byutilizing structure tensor field. Also to reduce the computational cost, we previously estimated nonchanging pixels and exclude them from calculation for speed up. Experiments shown that, proposed method performs better with low computational cost than general purpose TV on text image.

This paper presents a prior model for text image super-resolution in the Bayesian framework. In contrast to generic image super-resolution task, super-resolution of text images can be benefited from strong prior knowledge of the image class: firstly, low-resolution images are assumed to be generated from a high-resolution image by a sort of degradation which can be grasped through example pairs of the original and the corresponding degradation; secondly, text images are composed of two homogeneous regions, text and background regions. These properties were represented in a Markov random field (MRF) framework. Experiments showed that our model is more appropriate to text image super-resolution than the other prior models.