An improved image magnification algorithm for color images

Abstract

Preserving both edge and texture structures is a challenge to most magnification methods. In this paper an improved image magnification algorithm for color images is presented to cater for this challenge. A proposed estimation method to decide on a strong edge is utilized using a global threshold that has been determined automatically based on image statistics. Better performance compared to the original directional cubic convolution interpolation algorithm is achieved with no need for parameters settings. Simulation results demonstrate that our improved algorithm is able to faithfully reconstruct the magnified image, preserve edges and textures and reduce the common interpolation artifacts. Furthermore, it generates a higher visual quality of the magnified images and achieves higher peak signal to noise ratio (PSNR) and structural similarity (SSIM) compared to other methods.