一种基于核主成分分析的图像超分辨率算法

测试样本和训练样本集的匹配是基于学习的超分辨率算法中关键问题之一。本文方法通过将低分辨率的观察样本映射到高维的核空间中,实现测试样本和训练样本集的准确匹配,避免了基于学习的超分辨率算法中错误匹配问题,提高生成图像的质量。该算法包括:测试样本对训练样本集进行核主成分分析(kernel principal components analysis, KPCA); 利用距离约束算法得到在输入空间中的原像;最后将新生成的图像块进行重组,得到高分辨率的图像。在USPS数据集上进行的实验验证和对比分析表明：基于KPCA的图像超分辨率方法能够取得较好的超分辨率效果。

Kernel principal components analysis based super resolution method

The match between the observed example and the training example set is one of the crucial problems in learning based super resolution.The proposed method can make the match more accurate by mapping the observation example of low resolution to the reproducing kernel Hilbert space,avoiding the wrong match in the learning based super resolution and improving the image quality.The algorithm was first to apply KPCA to training examples to form a subspace,and then project the observed example onto the subspace.Th...