Filter Design for Orthogonal Two-Channel Filter Bank - Documentation

---

Go back to previous page.

A program for the design of a minimum phase lowpass FIR
analysis filter for a two-channel orthogonal filter bank
(wavelet system).

The filter has a specified ripple size and a specified
degree of flatness.

The filters produced by this program achieve a trade-off
between the classical Daubechies (Herrmann) filter and an
equi-ripple solution.

This permits a trade-off between frequency selectivity and
the number of vanishing moments.

The Daubechies (maximally flat) filter is obtained as a
special case by using L/2==N below.

program name : fir_orthog.m

subprogram   : oref.m, local_max.m, choose.m

________________________________________________________
help on fir_orthog _____________________________________

   [h,h2,h2] = fir_orthog(L,N,del)
   Design of a minimum phase lowpass filter for a two
   channel orthogonal filter bank (wavelet system).
   by : Ivan Selesnick, Rice University.
   input
     L   : length of filter (must be even)
     N   : degree of flatness
     del : ripple size (in magnitude squared)
   need L/2-N even and L/2 >= N
   output
     h : minimum phase wavelet filter
     h = conv(h2,h2); h2 contains the roots at z=-1,
                      h2 contains all the other roots.
   subprograms:
     oref.m, local_max.m, choose.m
   % EXAMPLE
        L = 14;
        N = 5;
        del = 0.01;
        [h,h2,h2] = fir_orthog(L,N,del);

________________________________________________________
    

Please report any bugs or send comments regarding the programs to selesi@ece.rice.edu