الفهرس 
abstractOnly 14 pages are availabe for public view
 |  | from | 99 |  |  |
| | | from | 99 | | |
Abstract
Image de-noising is one of the main steps in image analysis process. In order to obtain accurate image segmentation or recognition, images should pass through an efficient image de-noising technique. In medical imaging area, most of the images have some visual noise that results from the capturing machines such as the ultrasound machine. This will cause a great problem in diagnosing human cases if the diagnosis is not accurate. Hence, noise in medical images is considered a main and a complicated problem that needs an intelligent and efficient technique in order to be removed.
Although there are many current medical image de-noising techniques, most of them still need modifications to enhance their performances in denoising, especially the traditional techniques such as VisuShrink which depends on a very simple universal threshold equation to calculate either a hard or a soft threshold. This threshold is used in de-noising the images which may obtain good results in some images, but not for all images.
In this thesis, the intelligent classifiers were proposed depend on a set of efficient features to calculate the threshold. These features are extracted from the image using Scale Invariant-Feature Transform (SIFT) technique.
After that, they are used to train one of the introduced intelligent classifiers with the optimal threshold as a target. The trained technique could then be used to assign a suitable threshold to each new image based on its features.