基于ERBF核函数和边界填充的图像插值算法.pdfVIP

33 2 2007 1 Vol.33 No.2 Computer Engineering January 2007 ·· 2007 A TP391.41 ERBF 410004 ERBF ERBF 25%PSNR 33.81dBAMI ERBF Image Interpolation Algorithm Based on ERBF Kernel Function and Boundary Filling CHE Shengbing, HUANG Da (College of Electron Information, Central South Forestry University, Changsha 410004) AbstractBased on the relevancy among the pixels in the vicinity, expanding the pixels on the edge of the image, filling the expanding areas with the first row and column, making curve fit from known pixels by ERBF kernel function, it solves the problem dexterously of sampling pixels, that do not include the original image pixels. Experimentations suggest that, when the low resolving image, obtained by taking out 25% pixels with even rows and even columns from original image, become the high resolving image after interpolating, the parameter PSNR of the latter is 38.53dB, which is higher than any other image interpolation algorithms. And, there is no need to choose a threshold, which is always a difficult problem. It also puts forward a new effective method to compare different image interpolation algorithms by absolutely minus image for the first time. Key wordsERBF kernel function; Boundary filling; Image interpolation; Pixel fit; Absolutely minus image 1 [1] [2] ERBF (Exponential Radial Basis kernel Function) Baptista 1999 X Y [3] [4] Baptista