Title :
A family of wavelets which are dilatable by simple IIR filters [EEG analysis]
Author_Institution :
Dept. of Electr. Eng., Univ. of the Witwatersrand, Wits, South Africa
Abstract :
The family of wavelets called neural wavelets has zero mean, spectral peak at a frequency ∝√n and 20 n dB/decade rolloff where n is the order of the wavelet. The effective duration shortens with increasing n for a given t 0, where the parameter t0 allows time scaling of the wavelets. A simple infinite impulse response (IIR) filter is developed which expands or contracts the wavelet and can be designed to change the order n. The filter is used to detect partially overlapping constituent wavelets of a composite waveform. Filter design and noise sensitivity are discussed
Keywords :
digital filters; electroencephalography; filtering and prediction theory; medical signal processing; wavelet transforms; EEG analysis; IIR filter; design; dilation filter; infinite impulse response filter; neural evoked responses; neural wavelets; noise sensitivity; partially overlapping constituent wavelets; Africa; Electroencephalography; Electronic mail; Equations; Frequency; IIR filters; Tail; Wavelet analysis; Wavelet domain; Wavelet transforms;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1992. ICASSP-92., 1992 IEEE International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-0532-9
DOI :
10.1109/ICASSP.1992.226430
