Single-image super-resolution via local learning
- PDF / 1,301,953 Bytes
- 9 Pages / 595.276 x 790.866 pts Page_size
- 67 Downloads / 201 Views
ORIGINAL ARTICLE
Single-image super-resolution via local learning Yi Tang • Pingkun Yan • Yuan Yuan • Xuelong Li
Received: 23 November 2010 / Accepted: 16 January 2011 / Published online: 12 February 2011 Ó Springer-Verlag 2011
Abstract Nearest neighbor-based algorithms are popular in example-based super-resolution from a single image. The core idea behind such algorithms is that similar images are close in the sense of distance measurement. However, it is well known in the field of machine learning and statistical learning theory that the generalization of the nearest neighbor-based estimation is poor, when complex or high dimensional data are considered. To improve the power of the nearest neighbor-based algorithms in single-image based super-resolution, a local learning method is proposed in this paper. Similar to the nearest neighbor-based algorithms, a local training set is generated according to the similarity between the training samples and a given test sample. For super-resolving the given test sample, a local regression function is learned on the local training set. The generalization of nearest neighbor-based algorithms can be enhanced by the process of local regression. Based on such an idea, we propose a novel local-learning-based algorithm, where kernel ridge regression algorithm is used in local regression for its well generalization. Some experimental results verify the effectiveness and efficiency of the local learning algorithm in single-image based superresolution. Keywords Super-resolution Local learning Generalization Reproducing kernel Kernel ridge regression Similarity
Y. Tang P. Yan (&) Y. Yuan X. Li Center for OPTical IMagery Analysis and Learning (OPTIMAL), State Key Laboratory of Transient Optics and Photonics, Xi’an Institute of Optics and Precision Mechanics, Chinese Academy of Sciences, Xi’an 710119, Shaanxi, P. R. China e-mail: [email protected]
1 Introduction 1.1 Single-image super-resolution Super-resolution is a technique for enhancing the resolution of given images. Traditionally, such methods can be categorized as multi-image super-resolution and singleimage super-resolution based on the image input. Within the framework of multi-image super-resolution, a highresolution image will be generated using multiple low images of same scene, e.g. [1,2]. Different low-resolution image contains different details related to the high-resolution image. The focus of multi-image super-resolution is locating and fusing the information of details connected with high-resolution image. Within the framework of single-image super-resolution, high-resolution image is generated with just a single low-resolution image. It is clear that the single-image super-resolution is a more challenging problem in the field of super-resolution for no enough information on high-resolution details in hand. In order to make meaningful single-image super-resolution, prior information on high-resolution image is needed. The prior information can be presented directly or indirectly, which results in diff
Data Loading...
