Two-dimensional frequency-domain blind system identification using higher order statistics with application to texture synthesis

Authors
Citation
Cy. Chi et Ch. Chen, Two-dimensional frequency-domain blind system identification using higher order statistics with application to texture synthesis, IEEE SIGNAL, 49(4), 2001, pp. 864-877
Citations number
27
Language
INGLESE
art.tipo
Article
Categorie Soggetti
Eletrical & Eletronics Engineeing
Journal title
IEEE TRANSACTIONS ON SIGNAL PROCESSING
ISSN journal
1053-587X → ACNP
Volume
49
Issue
4
Year of publication
2001
Pages
864 - 877
Database
ISI
SICI code
1053-587X(200104)49:4<864:TFBSIU>2.0.ZU;2-V
Abstract
In this paper, Shalvi and Weinstein's super-exponential (SE) algorithm usin g higher order statistics for blind deconvolution of one-dimensional (1-D) linear time-invariant systems is extended to a two-dimensional (2-D) SE alg orithm. Then, a 2-D frequency-domain blind system identification (BSI) algo rithm for 2-D linear shift-invariant (LSI) systems using the computationall y efficient 2-D SE algorithm and the 2-D linear prediction error filter is proposed. In addition to the LSI system estimate, the proposed BSI algorith m also provides a minimum mean square error (MMSE) equalizer estimate and a n MMSE signal enhancement filter estimate. Then, a texture synthesis method (TSM) using the proposed BSI algorithm is presented. Some simulation resul ts to support the efficacy of the proposed BSI algorithm and some experimen tal results to support the efficacy of the proposed TSM are presented. Fina lly, some conclusions are drawn.