Jeng Hong Eng ¹, Azali Saudi ², Jumat Sulaiman ¹

Affiliations
1 Faculty of Science and Natural Resources, Universiti Malaysia Sabah, MY
2 Knowledge Technology Research Unit, Faculty of Computing and Informatics, Universiti Malaysia Sabah, 88400 Kota Kinabalu, Sabah, MY

Abstract

Objectives: This study examines the potential of Explicit Decoupled Group (EDG) iterative method to solve Poisson image blending problem in terms of blending iterations, time and the quality of newly blended images. Methods: The finite difference method is used to discretize the Poisson equation. Then the Poisson system is solved by EDG method via fivepoint rotated Laplacian operator. Three experiments are conducted to collect the findings obtained from classical Gauss- Seidel (GS), Explicit Group (EG) and EDG iterative methods. Findings: Numerical results showed that EDG is the most efficient iterative method in solving the proposed problem with the least composing time and number of iterations. This is due to the reason that the solution domain is halved, whereas the remaining can be computed directly. Meanwhile, the newly generated images by all the selected iterative methods are blended gracefully without obvious seam. Improvement: The EDG iterative method is better than other existing classical iterative methods in solving the Poison image blending problem.

Keywords

Explicit Decoupled Group Iteration, Finite Difference Scheme, Five-Point Rotated Laplacian Operator, Poisson Equation, Poisson Image Blending

Full Text