Title :
Two-dimensional two channel FIR signal adapted filter banks
Author :
Sheeba, V.S. ; Elias, Elizabeth
Author_Institution :
Dept. of Electron., Nat. Inst. of Technol., Calicut
Abstract :
In this paper the optimality results of two-dimensional, two-channel filter banks for different subsampling matrices were compared. It is seen that quincunx matrix is optimum in most of the natural images and also quincunx filter bank is a good match to the human visual system. A sub-optimal solution was also proposed that extends one-dimensional signal adapted filter bank design procedure to two-dimensional(2D) nonseparable FIR signal adapted filter banks using quincunx subsampling matrix. A cascade structure is used for the factorization of 2D paraunitary polyphase matrix and simulation of the nonseparable filter bank is done using iterative greedy algorithm. The simulated FIR responses match the ideal responses very closely
Keywords :
FIR filters; digital signal processing chips; digital signals; feature extraction; 2D paraunitary polyphase matrix; FIR signal; human visual system; iterative greedy algorithm; natural images; nonseparable filter bank; orthonormal filter banks; principal component filter banks; quincunx filter bank; quincunx matrix; subsampling matrices; two channel filter banks; two-dimensional filter bank; Data compression; Filter bank; Finite impulse response filter; Greedy algorithms; Humans; Multiresolution analysis; Signal design; Signal processing; Signal resolution; Visual system; Orthonormal filter banks; Principal component filter banks; Quincunx subsampling; Two-dimensional filter banks;
Conference_Titel :
Norchip Conference, 2006. 24th
Conference_Location :
Linkoping
Print_ISBN :
1-4244-0772-9
DOI :
10.1109/NORCHP.2006.329229