원문정보
초록
영어
In this paper, we present a new form of Fourier series named Kernel Fourier series (KFS) which produces time-frequency coefficients similar to wavelet. Both continuous and discrete forms of KFS are presented together with inverse KFS formulated to perform signal approximation. As KFS coefficients are dependent upon selection of kernel function, it can be used in different applications. Results from KFS test on feature extraction, signal analysis and signal estimation are presented and discussed.
목차
Abstract
1. Introduction
2. Preliminaries
2.1. Fourier Series
2.2. Kernel Methods
3. Kernel Fourier Series
3.1. Continuous Form of KFS
3.2. Kernel Fourier Series Properties
3.3. Discrete Form of KFS
3.4. Inverse KFS for Signal Estimation
3.5. KFS Applications
4. Experimental Results
4.1. KFS Coefficients of a Sine Function
4.2. KFS-Based Feature Extraction
4.3. Study of Influence of an Abrupt Change in Signal on KFS Coefficients
4.4. Noise Suppression
4.5. Signal estimation
5. Conclusion and Future Work
References
1. Introduction
2. Preliminaries
2.1. Fourier Series
2.2. Kernel Methods
3. Kernel Fourier Series
3.1. Continuous Form of KFS
3.2. Kernel Fourier Series Properties
3.3. Discrete Form of KFS
3.4. Inverse KFS for Signal Estimation
3.5. KFS Applications
4. Experimental Results
4.1. KFS Coefficients of a Sine Function
4.2. KFS-Based Feature Extraction
4.3. Study of Influence of an Abrupt Change in Signal on KFS Coefficients
4.4. Noise Suppression
4.5. Signal estimation
5. Conclusion and Future Work
References
저자정보
참고문헌
자료제공 : 네이버학술정보