Towards adaptive split spectrum processing

Author

Gustafsson, Mats G.

Author_Institution

Dept. of Technol., Uppsala Univ., Sweden

Volume

1

fYear

1995

Firstpage

729

Abstract

New results relevant for adaptive split spectrum processing (SSP) for ultrasonic clutter suppression are reported. Joint optimization of the filter bank and the nonlinearity is considered as well as how SSP can be considered performing maximum likelihood estimation. The joint optimization is realized by means of an artificial neural network which is trained using supervised learning like in system identification. The maximum likelihood approach is based on a matrix formulation of the filter bank processing

Keywords

acoustic filters; acoustic signal processing; adaptive signal processing; backscatter; clutter; flaw detection; learning (artificial intelligence); maximum likelihood estimation; multilayer perceptrons; spectral analysis; ultrasonic materials testing; ultrasonic scattering; adaptive split spectrum processing; artificial neural network; filter bank; joint optimization; matrix formulation; maximum likelihood estimation; multilayer perceptron neural network; supervised learning; system identification; ultrasonic backscattering; ultrasonic clutter suppression; Adaptive filters; Delay lines; Filter bank; Finite impulse response filter; Linear systems; Maximum likelihood estimation; Neural networks; Neurons; Supervised learning; System identification;

fLanguage

English

Publisher

ieee

Conference_Titel

Ultrasonics Symposium, 1995. Proceedings., 1995 IEEE

ISSN

1051-0117

Print_ISBN

0-7803-2940-6

Type

conf

DOI

10.1109/ULTSYM.1995.495672

Filename

495672