Digital image processing I

Abstract

Image enlargement is commonly used in digital image processing software. The classical method always has issues on visual artifacts such as edge blurring and bad visual quality of the image. The objective of my project is to propose an algorithm which can improve those problems. New image interpolation algorithms which preserve image edges are suggested in this report. The image is enlarged by an algorithm which interpolated by pixel data dependent triangulation. There are 3 different algorithm will be presented and compared. They are simple interpolation and interpolation with edge detection in triangulation in 3 X 3 squares and 16 X 16 squares. High resolution image was constructed by triangulation interpolation of low resolution image. The proposed method can is simpler and faster so it greatly improves computation efficiency in terms of the computation time and memory. The qualities of outcome images are compared visually and also by the Mean square error (MER) as assessment tool which gives a commutative error of the high resolution image from the low resolution image. The experimental result shows that the MER of the proposed algorithm was greatly minimized. The proposed algorithm was implemented by .NET C# in Visual Studio 2005.