November 16, 2024
Taher Yazdanpanah

Taher Yazdanpanah

Academic Rank: Associate professor
Address:
Degree: Ph.D in -
Phone: -
Faculty: Faculty of Intelligent Systems and Data Science

Research

Title
Spectral Theory of Polynomial Transforms and Fourier Transform on Graphs
Type Thesis
Keywords
Discrete Fourier Transform Eigenstructure Hermite-Gaussian functions Centered Discrete Fourier Tansform
Researchers fatemeh zarei (Student) , Taher Yazdanpanah (Primary advisor) , Ali Bagheri-Bardi (Primary advisor) , Milos Dakovic (Advisor)

Abstract

The Discrete Fourier Transform (DF T) is recognized as one of the key tools in digital data processing, such as data compression, dominant frequency identification, and audio and image processing. This research aims to provide a theoretical analysis of the eigenstructure of DF T and its related families, with the goal of introducing a new eigenbasis that enhances the efficiency of these transforms in signal processing. To achieve this, a novel and effective transformation system has been introduced, establishing a deep connection between the eigenstructure of the DF T and the eigenstructure of certain symmetric normalized discrete trigonometric transforms. This shift in approach led to the analysis of the eigenstructure of normalized discrete trigonometric transforms, through which a class of symmetric matrices that are square roots of the identity matrix, including symmetric normalized discrete trigonometric transforms, was introduced. Finally, utilizing advanced linear algebra techniques, a specific theoretical framework for extracting eigenvectors and performing spectral decomposition for this class was presented. Another significant achievement of this research is the introduction of an eigenbasis converging to Hermite-Gaussian functions, which play a key role in the spectral analysis of the continuous Fourier transform and have diverse applications in physics and electrical engineering. Although direct sampling from these functions cannot form a complete orthonormal basis for the DF T, this research, through innovative methods and precise design, introduces an eigenbasis for the centered discrete Fourier transform (CDF T) that, while simple, closely approximates Hermite-Gaussian functions. The results of the study indicate that leveraging the eigenbasis of DT T can improve the spectral analysis of DF T and have widespread applications in various fields. In conclusion, suggestions for expanding this research emphasize the importance of further exploring the