0

我正在尝试使用 10 点 daubechies 过滤器获取图像的第 4 级 DWT。（并自然地反转它！）

host = double(imread('lena512.bmp'));       % Load image data
nLevel = 4;                                 % Number of decompositions
cwA = cell(1,nLevel);                       % Approximation coefficients
cwH = cell(1,nLevel);                       % Horizontal detail coefficients
cwV = cell(1,nLevel);                       % Vertical detail coefficients
cwD = cell(1,nLevel);                       % Diagonal detail coefficients

% Do the DWT
myImage = host;
for iLevel = 1:nLevel,
  [cwA{iLevel},cwH{iLevel},cwV{iLevel},cwD{iLevel}] = dwt2(myImage,'db10');
  myImage = cwA{iLevel};
end

% Do the inverse DWT
fullRecon = cA{nLevel};
for iLevel = nLevel:-1:1
  fullRecon = idwt2(fullRecon,cH{iLevel},cV{iLevel},cD{iLevel},'db10');
end

上面的代码不断给我错误：

？？？错误使用 ==> plus Matrix 维度必须一致。

==> idwt2 在 93 x = upsconv2(a,{Lo_R,Lo_R},sx,dwtEXTM,shift)+ ... % 近似值时出错。

==> 18 时的虚拟错误 fullRecon = idwt2(fullRecon,cH{iLevel},cV{iLevel},cD{iLevel},'db10');

我相信这是由于反向 dwt 做了一些时髦的事情。我也尝试更改 dwtmode 但没有帮助。我真的很乐意提供任何帮助。

PS：lena512.bmp只是lena的灰度图。它的尺寸是 512x512。

我对新想法持开放态度 =)

4

1 回答 1

1

将代码重写为

clear all;
host = double(imread('lena512.bmp'));       % Load image data
nLevel = 4;                                 % Number of decompositions
cwA = cell(1,nLevel);                       % Approximation coefficients
cwH = cell(1,nLevel);                       % Horizontal detail coefficients
cwV = cell(1,nLevel);                       % Vertical detail coefficients
cwD = cell(1,nLevel);                       % Diagonal detail coefficients

% Size matrix
s = [size(host,1) size(host,2)];

% Do the DWT
myImage = host ;
for iLevel = 1:nLevel
  [cwA{iLevel},cwH{iLevel},cwV{iLevel},cwD{iLevel}] = dwt2(myImage, 'db10');
  s = [s; size(cwH{iLevel},1) size(cwH{iLevel},2)];
  myImage = cwA{iLevel};
end

% Do the inverse DWT
fullRecon = cwA{nLevel};
for iLevel = nLevel:-1:1
  fullRecon = idwt2(fullRecon,cwH{iLevel},cwV{iLevel},cwD{iLevel},'db10',s(iLevel,:));
end

解决了我的问题。希望它可以帮助其他人...

于 2011-06-12T22:36:34.377 回答