摘要： 研究核回归图像插值问题，基于各向异性距离方法，提出一种边缘保持的核回归图像插值方法。分析不同参数对该方法以及 Steering核回归方法的影响。实验结果表明，边缘保持的核回归图像插值方法对于参数变化具有较强的稳定性，算法鲁棒性较高，插值图像的边缘信息也得到更好的保持。

关键词: 图像插值, 核回归, 各向异性, Steering核, 边缘保持

Abstract: This paper studies the kernel regression for image interpolation. Basing on anisotropic distance method, it proposes an edge-preserved kernel regression image interpolation method. It analyzes the influence of parameters to the proposed method and the Steering method. Experimental results show that the proposed method has strong stability and high robustness to variation of parameters, and very successful in preserving edges.

Key words: image interpolation, kernel regression, anisotropic, Steering kernel, edge-preserved

中图分类号: 

• TP391.41