Super Ellipse Window Based Image Zooming Algorithm

Abstract

Image zooming deals with the enlargement of an image. While zooming an image, an image is magnified but it introduces certain artifacts. These artifacts are jaggies, blurring and ghosting. So while zooming an image these artifacts should be reduced. In this thesis, Introduction gives a clear picture about the type of image processing, different formats of images along with the brief introduction about image zooming and its interpolation techniques. It also compares the interpolation techniques on the basis of visual and computational properties. Literature survey gives the details of the category of algorithms i.e. adaptive algorithm and non-adaptive algorithm with detailed description of Pixel replication, Bilinear interpolation, Bicubic interpolation, Filtering methods, Learning based algorithms. It also describes the literature survey of the image zooming interpolation techniques. Image causes a lot of problems when it is zoomed so in order to reduce these artifacts and produce sharp images an algorithm is proposed along with the details of two existing algorithms i.e. New Edge Directed Interpolation (NEDI) and Improved New Edge Directed Interpolation (iNEDI). The proposed algorithm provides an improvement over the existing algorithms which was visible through images and also through a quantitative measure known as Peak Signal to Noise Ratio (PSNR). The proposed algorithm provides maximum PSNR improvement of 0.8 dB over NEDI and 0.7 dB over iNEDI.