Data fusion of very high resolution satellite images for enviromental application

مقدم أطروحة جامعية

Ali, Mahdi Mustafa

الجامعة

الجامعة التكنولوجية

الكلية

-

القسم الأكاديمي

قسم هندسة البناء و الإنشاءات

دولة الجامعة

العراق

الدرجة العلمية

ماجستير

تاريخ الدرجة العلمية

2008

الملخص الإنجليزي

The accuracy and interpretability of VHR satellite Quick Bird data make it ideal data for new city planning, urban growth studies, and environment studies.

In the present study, two multitemporal Quick Bird images (Pan and MS) are available.

These images covered the study area which represents the central part of Baghdad city.

The geometric correction for available images has been done using two dimensional (2D) 1st and 2nd order polynomials ; nonparametric model with 2K GPS-UCPs.

Necessary Image Enhancement techniques such as linear contrast stretch, histogram matching, and histogram equalizalion are performed.

The main objective of (his study is to use image fusion techniques to obtain information of high quality which may consist of more accurate description of the scene than any of the individual source images.

For this purpose, Quick Bird data have heen merged using nine methods namely ; PC A, ML, BT, WT-PC, WT-HSI, HIS, OS, CN-ttr, and HSV.

Statistical and numerical comparisons have heen made (relative shift in hand correlation matrix.

mean and standard deviation) to evaluate the effect of different fusion techniques on distortion of spectral characteristics of higher spectral resolution data.

It is found that the PC A method is the most effective in preserving the spectral information contained in the original MS image-Initial and final classification, supervised and unsupervised, is applied to the original MS image and fused image respectively.

A comparison Lo the results is done indicating that the unclassified pixels from the initial classification are totally classified in the final classification carried out on the fused image and then improving the classification accuracy (the 2nd objective), producing more accurate thematic land use / land cover map of six.

classes.

Overall Classification accuracy is 90 %.

All of the study results are supported with necessary tables, graphs and band correlation coefficient matrices.

التخصصات الرئيسية

الهندسة المدنية
تكنولوجيا المعلومات وعلم الحاسوب

الموضوعات

• التخطيط الحضري
• الجيولوجيا
• الأقمار الصناعية
• العراق

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-305554