Title :
Time-varying filter banks and wavelets
Author :
Sodagar, Iraj ; Nayebi, Kambiz ; Barnwell, Thomas P., III
Author_Institution :
Sch. of Electr. Eng., Georgia Inst. of Technol., Atlanta, GA, USA
fDate :
11/1/1994 12:00:00 AM
Abstract :
Time-varying filter banks and wavelets are studied and a design procedure is presented. In the resulting analysis-synthesis structures, the analysis filters and the corresponding synthesis filters, the number of bands, and the decimation rates can be changed with time. Such structures can be considered as time-frequency overlapping block transforms. From this viewpoint, the tiling of the time-frequency plane and the corresponding basis functions are changed in time. The time-varying discrete wavelet transforms can be considered a special case of time-varying overlapping block transforms and are studied in detail. The formulation is based on the time domain formulation of time-varying analysis-synthesis structures. The design procedure can be used to design time-varying perfectly invertible transformations with a finite number of distinct analysis structures. For adaptive filter bank application, a least squares design method is also studied
Keywords :
adaptive filters; digital filters; filtering theory; least squares approximations; time-domain analysis; time-varying filters; wavelet transforms; adaptive filter bank; analysis filters; analysis-synthesis structures; basis functions; decimation rates; least squares design method; perfectly invertible transformations; synthesis filters; time domain formulation; time-frequency overlapping block transforms; time-frequency plane; time-varying discrete wavelet transforms; time-varying filter banks; wavelets; Channel bank filters; Discrete wavelet transforms; Filter bank; Fourier transforms; Image reconstruction; Karhunen-Loeve transforms; Time domain analysis; Time frequency analysis; Time varying systems; Wavelet transforms;
Journal_Title :
Signal Processing, IEEE Transactions on