Error- Subscript indices must either be real positive integers or logicals.

I have a code for dual tree wavelet transform

x = rand(64,64,3);
  J = 3;
  [Faf, Fsf] = FSfarras;
  [af, sf] = dualfilt1;
  w = dualtree3D(x, J, Faf, af);
  y = idualtree3D(w, J, Fsf, sf);
  err = x - y;
  max(max(max(abs(err))))

for this i get the answer ,

but if i replace x by my image 

x=imread('st.tif'), i get error,

please help

[Walter Roberson]

Please show the error traceback.

Also please show class(x) after the imread. Not double precision, right, whereas your x = rand() is initializing to double precision. Consider whether you should use im2double(x)

Answer 1

[Wayne King]

A couple things:

1.) Is this an RGB image? The 3-D wavelet transform is not suitable for an RGB image. The 3-D wavelet transform is intended for actual 3-D data. An RGB image is 2-D data and you should use a 2-D wavelet transform that accepts RGB image inputs, like the routines in the MathWorks' Wavelet Toolbox.

2.) Even if your data is suitable for the 3-D wavelet transform, what is the size of the third dimension? Professor Selesnick routines are not set up to handle dimensions that are not divisible by 2. So for example:

 x = randn(64,64,63);
 J = 3;
 [Faf, Fsf] = FSfarras;
 [af, sf] = dualfilt1;
 w = dualtree3D(x, J, Faf, af);

will produce the exact error you see, but

 x = randn(64,64,10);

will not.