提高图像轮廓清晰度的快速插值放大方法  

Fast Image Interpolation Algorithms Based on Improved Definition of Profiles on Images

作　　者：谢征[1] 李莉[1] 李玉峰[1] 沈春林[1] 

机构地区：[1]南京航空航天大学自动化学院,南京210016

出　　处：《南京航空航天大学学报》2007年第2期187-192,共6页Journal of Nanjing University of Aeronautics & Astronautics

基　　金：教育部博士点基金(20030287008)资助项目

摘　　要：分析了当前主要的高清晰度图像插值放大方法,为实现快速放大而选择了基于图像轮廓插值改善清晰度的技术路线,并在已有算法的基础上开展了进一步的研究。提出了新的图像轮廓线光顺数值算法,对轮廓线划分的两类子区域提出了精度更高的插值算法。最后,以典型平面图形的图像和由3DM ax生成的三维渲染图像为例对本文方法进行检验,与双线性插值结果的对比说明,该方法可以明显改善插值放大效果。Fast image magnification methods can keep a high resolution, thus they are very important in real-time display of 3D movies. The interpolation based on the refinement of shape-pr0files is adopted for image magnifications as fast as possible, by analyzing main advanced methods of image magnifications. To improve this method, a new algorithm for smoothness shape-profiles is studied by analogy the thermal diffusion. Moreover, the interpolation based on such smoothed profiles is improved with the higher accuracy. Image magnification cases including images after 3D rendering by 3DMax, show that the new method can quickly obtain magnified images with abviously better quality than Bi-linear interpolation.

关 键 词：图像放大 图像轮廓 快速插值 光顺 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]