The following information was submitted:
Transactions: INTERNATIONAL JOURNAL of CIRCUITS, SYSTEMS and SIGNAL PROCESSING
Transactions ID Number: 19-196
Full Name: Majid Mohammady Oskouei
Position: Assistant Professor
Age: ON
Sex: Male
Address: Mining Engineering Faculty- Sahand University of Technology- Tabriz
Country: IRAN
Tel: 09144197873
Tel prefix: +98
Fax: 04123459299
E-mail address: moskouie@yahoo.com
Other E-mails: mohammady@sut.ac.ir
Title of the Paper: Detection of the number of signal sources in hyperspectral data
Authors as they appear in the Paper: Majid Mohamady Oskouei, Rashed Poormirzaee
Email addresses of all the authors: moskouie@yahoo.com, Rashed.poormirzaee@gmail.com
Number of paper pages: 8
Abstract: This study investigates and compare different methods to estimate the number of signal sources in the hyperspectral data. To achieve an accurate map of mineral distributions in the study area by means of the spectral analysis of Hyperion data, the number of endmembers was computed by different methods. This process is also known as determination of virtual dimensionality of the image. Estimation of Virtual Dimensionality of data or in other words, the number of detectable endmembers from data is very important task in hyperspectral imagery. If this estimated number doesn't meet the reality, final estimation of mineral abundances will be erroneous. The results established that principle component analysis underestimates the virtual dimensionality of data. This is reasonably due to lower abundances of some minerals on the earth surface that will be considered as unimportant principle components because of their lower energy fraction in the total radiance measured at!
sensor. The higher order statistical method on the other hand, showed better performance. This method uses Neyman–Pearson detection theory and its estimation is more realistic.
Keywords: Hyperion, Hyperspectral imagery, Principle component analysis, Signal processing, Virtual dimensionality
EXTENSION of the file: .doc
Special (Invited) Session: Estimation of virtual dimensionality of hyperspectral data by principle component analysis and higher order statistical method
Organizer of the Session: 626-331
How Did you learn about congress:
IP ADDRESS: 80.191.211.36
