基于细分的图像插值算法  

A Subdivision Approach to Image Interpolation

作　　者：罗立彦[1] 杨勋年[1] 

机构地区：[1]浙江大学数学系图形图像研究所,杭州310027

出　　处：《计算机辅助设计与图形学学报》2006年第9期1311-1316,共6页Journal of Computer-Aided Design & Computer Graphics

基　　金：国家自然科学基金(60303015;60333010);国家重点基础研究发展规划项目(2002CB312101)

摘　　要：将基于法向的曲线细分技术应用到图像插值中,提出一种基于细分的图像插值算法.该算法无须建立中间连续图像模型,能自适应地插值,而且插值系数可为任意正实数.应用该算法插值后的边界清晰、自然,忠实地反映了原始图像的面貌.与传统的插值算法相比,其边界处理效果好,具有线性复杂度且易于实现.By extending the normal based subdivision scheme for curve and surface design, a new algorithm for image interpolation is proposed in this paper. The images will be zoomed in by interpolatory subdivision and no intermediate continuous surface model should be established. With proper choices of the subdivision parameters, images can be interpolated with any positive real scale while edges and borders within images can be kept well. Examples and comparisons with some other image interpolation methods show that the new image interpolation algorithm has the advantage of good edge preservation, linear complexity and easy implementation.

关 键 词：图像插值 图像处理 细分 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]