基于支持向量回归的图像超分辨率重建算法

文章提出一种新的基于支持向量回归(SVR)和稀疏表示的图像超分辨重建算法。SVR对输入数据有良好预测输出类别能力。图像统计表明,图像块可以从过完备字典中通过稀疏线性组合很好的表示。对一幅低分辨率输入图像,可以将图像超分辨问题视为在高分辨图像中估计其像素位置。与传统的支持向量回归方法相比,本文采用的特征是不同类型的图像块的稀疏表示。研究表明,稀疏表示作为特征对噪声有一定的鲁棒性。实验结果表明,本文方法与传统支持向量回归方法相比在图像重建质量上有一定的优势。

An Image Super Resolution Reconstruction Algorithm Based on Support Vector Regression

In this paper, a new approach to single-image Super Resolution（SR） based on support vector regression （SVR） with sparse representation is presented. SVR is known to offer excellent generalization ability in predicting output class labels for input data. The research on image statistics suggests that image patches can be well-represented as a sparse linear combination of elements from an over-complete dictionary. For a low resolution image input, we approach the SR problem as the estimation of pixel labels in its high resolution version. Compared with general SVR methods, the feature considered in this work is the sparse representation of different types of image patches. Prior studies have shown that this feature is robust to noise in image data. Experimental results show that our method is competitive or even superior in quality to images produced by conventional SVR method.