Date of Award


Degree Type


Degree Name

Master of Science


Computer Science


Dr. Mahmoud R. El-Sakka


Diffusion filters are designed to smooth homogenous image regions reducing noise and preserving edges. However, with the increase in the amount of diffusion applied over time, the smoothing effect over homogenous regions might not stop at its boundaries, leading to a blurring effect. This effect broadens image features’ boundaries and dislocates their edges. This thesis presents two attempts to correct the boundary broadening and features distortion drawbacks of diffusion filtering and hence, reaching a nearly stable diffusion over time. The work is based on tracking of prominent image features throughout the diffusion. The introduced stable diffusion algorithms are evaluated in terms of features preservation, edge localization and noise reduction. Two figure of merits, namely Pratt’s and Berkley segmentation benchmark, have been utilized to measure the accuracy of the diffusion algorithms in terms of edges localization. The results show that the diffused images and their extracted edge maps exhibit significant noise reduction with well localized edges and features preservation. The introduced algorithms have a better noise reduction capability compared to traditional de-noising filters.



To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.