几种插值算法的比较研究-《计算机技术与发展》

文章信息/Info

Author(s):: ZHANG Hai-yan; WU Fang; Zhuhai Campus of Beijing Institute of Technology

摘要:: 在图像放大处理中,一般都会用到插值算法,为了实时性的要求,在保证放大效果的基础上,希望插值算法尽可能简单。插值指利用某一个函数来计算出2个或更多的值之间的值。文中就常用的NEDI、DUBE、9-Direction、KR（kernel re-gression）等的理论基础和算法原理做了分析,通过采用四种插值算法对图像进行缩放处理操作,可以直观比较它们处理后的效果。文中使用matlab仿真工具分别就这四种算法对典型的lena、flower和night图像进行了处理,并对它们的处理结果加以比较,最后总结了四种算法各自的特点

Abstract:: Interpolation algorithm is used generally in real-time image upscale processing,to make sure image quality after upscale,interpolation algorithm should not be too much complex.Interpolation refers to a particular function to calculate the value of two or more values.It analyzes the theoretical basis and algorithms principle for the four interpolation algorithms：NEDI,DUBE,9-Direction,KR（kernel regression）,by using four kinds of interpolation algorithm for scaling the image processing operations can directly compare the effect of treatment.Matlab simulation tools were used in this article on the four algorithms on a typical lena,flower and night images were processed and to compare the result correspond to interpolation algorithm,finally summarize the complexity of four interpolation algorithms and image effect after upscale