DocumentCode :
1725169
Title :
A class of selective band-pass CNN linear filters
Author :
Matei, Radu P.
Author_Institution :
Tech. Univ. "Gh.Asachi" of Iasi, Romania
fYear :
2005
Firstpage :
341
Lastpage :
344
Abstract :
In this paper we introduce a design method for a class of linear spatial filters implemented on CNNs. The aim is to realize selective band-pass filters with templates of minimum-size. The design procedure relies on 1D HR low-pass and band-pass prototype filters, from which we realize different types of 2D band-pass filters, basically belonging to two classes, namely filters with circular symmetry (based on a LP-BP transformation) and separable filters. Simulation results are provided which illustrate the capabilities of these filters on a real grayscale image.
Keywords :
IIR filters; band-pass filters; cellular neural nets; low-pass filters; spatial filters; two-dimensional digital filters; 2D band-pass filters; CNN linear filters; LP-BP transformation; cellular neural nets; circular symmetry; infinite impulse response filters; linear spatial filters; selective band-pass filter; Band pass filters; Cellular neural networks; Centralized control; Cutoff frequency; Design methodology; IIR filters; Nonlinear filters; Prototypes; Spatial filters; Transfer functions;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
IEEE-NEWCAS Conference, 2005. The 3rd International
Print_ISBN :
0-7803-8934-4
Type :
conf
DOI :
10.1109/NEWCAS.2005.1496732
Filename :
1496732
Link To Document :
بازگشت