Image super-resolution using multi-layer support vector regression

Existing support vector regression (SVR) based image superresolution (SR) methods always utilize single layer SVR model to reconstruct source image, which are incapable of restoring the details and reduce the reconstruction quality. In this paper, we present a novel image SR approach, where a multi-layer SVR model is adopted to describe the relationship between the low resolution (LR) image patches and the corresponding high resolution (HR) ones. Besides, considering the diverse content in the image, we introduce pixel-wise classification to divide pixels into different classes, such as horizontal edges, vertical edges and smooth areas, which is more conductive to highlight the local characteristics of the image. Moreover, the input elements to each SVR model are weighted respectively according to their corresponding output pixel´s space positions in the HR image. Experimental results show that, compared with several other learning-based SR algorithms, our method gains high-quality performance.