This is where navigation should be.


GSP_DESIGN_CAN_DUAL - This function return the canonical dual filters of g

Usage

gd = gsp_design_can_dual( g );

Description

Inputs parameters:
g : cell array of filters tol : tolerance for the pseudo-inverse
Ouputs parameters:
g : cell array of filters

This function returns the canonical dual filterbank g. Note that it might not be the be the optimal solution in term of computation.

Example:

N = 100;
G = gsp_sensor(N);
G = gsp_compute_fourier_basis(G);
g = gsp_design_abspline(G,8);
gd = gsp_design_can_dual(g);
paramplot.show_sum = 0;
figure(1)
gsp_plot_filter(G,g,paramplot);
title('Original filters')
figure(2)
gsp_plot_filter(G,gd,paramplot);
title('Canonical dual filters');

x = rand(N,1);
param.method = 'exact';
coeff = gsp_filter_analysis(G,g,x,param);
xs = gsp_filter_synthesis(G,gd,coeff,param);
norm(xs-x)

This code produces the following output:

ans =

   1.1674e-14
gsp_design_can_dual_1_1.png gsp_design_can_dual_1_2.png