Nurul Afiqah Basran (2019) Analisis prestasi famili kaedah lelaran blok berpemberat dengan pengoperasi laplacian lima dan sembilan titik ke atas pengaburan imej. Masters thesis, Universiti Malaysia Sabah.
![[img]](https://eprints.ums.edu.my/style/images/fileicons/text.png) |
Text
24 PAGES.pdf Download (753kB) |
|
Text
FULLTEXT.pdf Restricted to Registered users only Download (25MB) |
Abstract
The image blurring process is one of the filtration techniques used in image processing field. The concept used in this study is to blur the image while maintaining the important features such as sharp edges where this study is ve,y crucial in the computer vision area. Accordingly, the motivation of this study is to apply the block iterative method on image blurring application through the solution of nonlinear diffusion equation. This is because the solution using the classic iterative method by the past researcher requires a high cost of computation in terms of processing time. In order to develop numerical methods to obtain solutions that are more efficient and faster by minimizing the cost calculations, implicit finite difference approximation of nonlinear diffusion equations is constructed via five and nine points Laplacian operator. Both operators are considered to issue corresponding approximation equations and apply to the desired linear system. Both operators are taken into account in deriving the corresponding approximation equations and applied to obtain a linear system which is then solved using considered iterative method. Therefore, this study attempts to verify the computational efficiency family of SOR point iterative method and family of block SOR point iterative method in solving finite difference approximation equations of nonlinear diffusion equations for the application of image blurring via five-point (SL) and nine-point (9L) Laplacian operators. There are four colour experimental images of sizes 256x256, 512x512 and 1024x1024 selected to test the efficiency of the calculations for the point and block iterative methods such as Gauss-Seidel (GS) Successive Over-Relaxation (SOR), two-point Explidt Group (2- EG), two-point Explicit Group Successive Over-Relaxation (2-ESOR), four-point Explicit Group (4-EG) and four-point Explicit Group Successive Over-Relaxation (4- EGSOR) where each of which was formulated based on five and nine point Laplacian operators. method is more efficient in reducing the number and time of iteration in solving nonlinear diffusion equations in image blurring compared with the control method (Jacobi-SL iteration method) at 81.00% - 86.75% and 73.77% -89.24% followed by 4-EGSOR-9L at 85.00% -89.50% and 65.58% -87.37%, respectively. The results of the blurred images of all the iteration methods showed no notable differences based on subjective and objective approach of image quality assessments with an SSIM index of greater than 0.9 which is close to one. Thus, this study can contribute to the field of imaging through the reduction of computational complexity in the image filtering process.
| Item Type: | Thesis (Masters) |
|---|---|
| Keyword: | Filtration techniques , Image , Laplacian operators |
| Subjects: | T Technology > TA Engineering (General). Civil engineering (General) > TA1-2040 Engineering (General). Civil engineering (General) > TA1501-1820 Applied optics. Photonics |
| Department: | FACULTY > Faculty of Science and Natural Resources |
| Depositing User: | DG MASNIAH AHMAD - |
| Date Deposited: | 28 Oct 2024 16:03 |
| Last Modified: | 28 Oct 2024 16:03 |
| URI: | https://eprints.ums.edu.my/id/eprint/41363 |
Actions (login required)
 |
View Item |