DBlockToolbox.zip_DBlockToolbox_反褶积

The Matlab Curvelet Toolbox

Beta version

Laurent Demanet

August 31, 2004

This code is not for distribution yet.

1 Setup

Copy the archive file CurveletToolbox.tar in the directory of your choice, and expand

it with the linux command

tar xvf CurveletToolbox.tar

Donoho (2004).

• Numerically tight up to machine precision: as a result the pseudo-inverse is sim-

ply the adjoint.

• The algorithmic complexity is O(N logN).

• Comes as a complex-valued or real-valued transform.

• The coarsest scale subband consists of Meyer wavelets, all the other subbands

consist of angularly selective curvelets.

• The redundancy is about 5 to 10.

• The input image can be a rectangle with arbitrary height and length.

1

‘wrapping’ technique. In particular, the cartesian grid is used throughout the computa-

tion and no interpolation is necessary in the frequency plane.

(All this will be expanded elsewhere)

3 How to use it

function C = FCT(x, is_real, nbscales, nbangles_coarse)

% FCT.m - Fast Curvelet Transform

%

% Inputs

% x M-by-N matrix

%

% nbscales number of scales including the coarsest wavelet level

% [default set to ceil(log2(min(M,N))/2)]

% nbangles_coarse

% number of angles at the 2nd coarsest level, minimum 8,

% must be a multiple of 4. [default set to 16]

%

% Outputs

% C Cell array of curvelet coefficients: C{j}{l}(k1,k2) is

% the coefficient at

% - scale j: integer, from finest to coarsest scale,

2