Tu-Satellite image analysis and quality estimation.+CD

number: 
2250
إنجليزية
department: 
Degree: 
Imprint: 
Computer Science
Author: 
Esraa Hussain Ali
Supervisor: 
Dr. Ali Abid Dawood Al-Zuky
year: 
2009

Abstract:

The image quality analysis techniques of TV-Satellite images come from the significant processing of high importance of image information in the multimedia world. In this research, the quality of TV-Satellite images on the three satellites (Arabsat, Hotbird and Nilesat) was studied to estimate the best quality image. A number of image quality metrics have been introduced. The introduced image quality metrics consist of two methods. The first method is called the contrast of image edges and the second method is called the cross correlation. In the contrast of image edges method, the edges have been extracted from the image by applying Sobel edge detector on the R, G, B and L components using different threshold values. The amount, the SNR, the mean and the standard deviation of image edges were calculated for all R, G, B and L components. The contrast of image edges metric had been determined for different threshold values using Michelson formula. The Michelson contrast equation was modified in different ways to present number of contrast approaches that applied on the edges of the image. In the cross correlation method the correlation of selected homogenous regions in different locations of the image has been computed for all R, G, B and L components. The regions' selection was made manually and automatically from the image. The results for the quality metrics used in this research indicated that TV-Satellite images from Nilesat hold better quality from Arabsat and Hotbird images. The quality of the images on the Arabsat came in the second place and the Hotbird became last.