摘要
提出了一种基于再生核的图像插值方法。再生核源于不同学科分支,已成为函数逼近的重要工具。该方法将再生核的再生公式离散,并按照证明的再生核数值积分方法导出了新型的图像插值算法。新插值算法的优点是利用再生核的数学模型特点保持图像的边界信息和光滑性,促进了插值的整体性能。实验结果表明:该算法能够克服其它算法的缺点,获取的高分辨率图像既能够保持图像的边界信息,又能保证图像的光滑性。
In this paper a novel image interpolation algorithm based on the reproducing kernel space W is proposed. Reproducing kernel developed in different disciplines has become an important tool for function approximation. The new method discretizes reproducing kernel and obtains interpolation formula with the proof of discrete integral. The advantages of the new interpolation method are the ability to keep the image edges and smoothness by model prop- erties of the reproducing kernel. Through comparisons with other algorithms it is shown that the new interpolation pro- duce higher quality, but visually it is very efficient at reducing jagged edges.
出处
《计算机仿真》
CSCD
2007年第3期219-222,245,共5页
Computer Simulation
基金
国家自然科学基金资助项目(60375007)
北京市自然科学基金资助(40410031)
北京市自然科学基金项目
北京市教育委员会科技发展计划重点项目(KZ200310005002)
多媒体与智能软件技术北京市重点实验室资助(KB10200584)
关键词
图像处理
图像插值
逼近方法
再生核
Image processing
Image interpolation
Approximation method
Reproducing kernel
