Effective model construction for enhanced prediction in example-based super-resolution

Single-image super-resolution is widely adopted for high resolution display related applications. Example learning-based approaches can provide plenty of image details by using trained dataset. Regression-based methods reduce the memory storage size by training mapping functions instead of using a huge dictionary. The reconstructed image quality can be further enhanced by combining various prediction results. This work presents an effective model reconstruction method for enhanced predictions. The desired model can be constructed offline when using the local multi-gradient level pattern as the clustering feature. Applying the proposed schemes can further improve the quality of reconstructed high resolution image while retaining almost the same time complexity as the original solution. Experimental results exhibit that the quality of reconstructed image using the proposed schemes is very close to that of Yang's work, but the proposed one can operate much faster than his solutions. Moreover, the space for storing mapping functions can be dramatically reduced by using the proposed model combining method.

I. Introduction

Single-image super-resolution (SISR) is attracting more and more attention nowadays because of the trends of high resolution display. Constructing an artifact-free high-resolution (HR) image from only one single low-resolution (LR) image is a critical and challenging task in many applications. Example learning-based SISR restores the details by using trained dataset. The co-occurrence of LR-HR patches [2] involved in the dataset would be helpful to learn the correlation between LR and HR patches for constructing a LR-to-HR dictionary. Since the dictionary would be very large for ensuring the quality of reconstructed images, the resulting storage requirement and speed limitation pose severe restrictions on real-time applications.