Digital video enhancement and denoising

Dissertant

abd al-Ghani, Fatin Zakariyya

University

University of Technology

Faculty

-

Department

Department of Electrical Engineering

University Country

Iraq

Degree

Master

Degree Date

2006

English Abstract

This thesis focuses on point-wise degradations, i.e., considering noise in video sequences.

White Gaussian additive noise is a commonly standard applied model to represent the noise as obtained in video signals under weak antenna signal conditions and AM (VSB) modulation.

After capturing noisy video scenes and extracting frames from these video files by Adobe after Effect software, the proposed algorithms enhance and deny the extracting frames.

A noise removal method have been proposed for color image sequences named by (3ADENCs), based on coupled spatial and temporal anisotropic diffusions.

In both cases, diffusion speed is determined by medium-specific (space or time) discontinuities detector.

Spatial discontinuities are measured via luminous gradient norm and it is important in stopping diffusion in edge area; while an (adaptive motion detection) was propose to detect temporal discontinuities and it is used in stopping diffusion in motion area.

The main advantages of this method is that it allows PDE formalism to provide fast yet stable algorithms and removes noise efficiently with high PSNR value.

To complete this work, a developed technique for spatiotemporal noise level estimation was proposed, that is more accurate and the least sensitive than varying noise levels and image sequence content among others.

Main Subjects

Information Technology and Computer Science

Topics

• Noise
• Data processing

• APA
• MLA
• AMA

Language

English

Data Type

Arab Theses

Record ID

BIM-305887