| 要約: | Image is normally contaminated by noise during its acquisition process. In order to deal with this problem, image denoising is introduced as the preliminary process in image processing. The approach of performing image denoising is currently an active research area as the techniques used frequently have some advantages or disadvantages, depending on the quality of the image used. The commonly used spatial filters such as median, mean and Gaussian filters would cause blurring effect on image. Therefore, a PDE-based image denoising will be introduced to overcome the problem. The technique is selected because it has a good denoising effect and hence, a better edge preservation on image. In this paper, a second-order PDE and a fourth-order PDE are computationally solve using finite difference method and implemented on radiographic images for the purpose of removing image noise. Then, the performances of the models are evaluated using peak signal to noise ratio (PSNR) to evaluate the image quality. In addition, the blocking artifacts of the processed image are measured using Tchebichef moment. Results have shown that the fourth-order PDE performed better in terms of image quality and it is able to reduce the blocking artifacts on noisy images as compared to the second-order PDE. © 2016 Author(s).
|
|---|