EEG data a two-dimensional matrix as 868 by 16, when it details coefficients are calculated using Matlab code the matrix of D1, D2 and D3 are different from when similar data is decomposed using wavelet Analyzer. Can any one guide me what dimension of D1,D2 one can get when decomposed a 868 by 16 matrix with DB3 of level 6
Your question is not presented correctly. You should put the code you are using to obtain the coefficients, and also explain what is "wavelet Analyzer", I assume is a different software than Matlab. Also, when you say that the matrix obtained are "different", you need to clarify what is different. Is the size of the matrix diferent? or the size is the same but the values are different?. Also you should specify which wavelet transform are you using.
Some tips about things you need to check:
- which wavelet transform are you using? it is not the same to use swt() or wavedec(), maybe you are mixing different transforms.
- You should check the extension mode for the transform. It is not the same to use periodization, that use symmetry or other available methods of extension. Periodization produces outputs with the lowest number of coefficients, other extension methods produce larger number of coefficients. If there is a size difference in your output, it may be produced by different extension modes. In matlab you can use dwtmode() to check which extension mode is used.
Thank you so much sir for you time .....I have a cell array comprises of three cell with dimension of 868 by 16 of each cell. When i apply db3 with level 6 on each cell and safe detailed coefficient and approximate coefficient in an array corresponding to original data cells by using Matlab code shown below:
dwt_coeffs_struct = struct('A', cell(size(data)), 'D', cell(size(data)));
% Loop over each cell in the data array
for i = 1:numel(data)
% Convert the cell data to double and reshape it to a column vector
signal = reshape(cell2mat(data{i}), [], 1);
% Initialize arrays to store approximation coefficients and detail coefficients
A_coeffs = cell(1, 6);
D_coeffs = cell(1, 6);
% Perform the DWT for each level
for j = 1:6
[C, L] = wavedec(signal, j, 'db3');
% Store the approximation coefficients and detail coefficients separately
if j == 1
A_coeffs{j} = appcoef(C, L, 'db3', j);
end
D_coeffs{j} = detcoef(C, L, j);
end
% Store the DWT coefficients for this signal in the structure array
dwt_coeffs_struct(i).A = A_coeffs;
dwt_coeffs_struct(i).D = D_coeffs;
end
Than i got detiled cofficents D1=41666 by 1
D2=20835 by 1
D3=10420 by 1
D4=5212 by 1
D5=2608 by 1
D6=1306 by 1
Am i getting correct dimensions of these detailed coffinites ....Please guide so that i may further extract features from these detailed coefficients
You are using a decimated wavelet transform, which is implemented using recursively a stage of filtering followed by downsampling. If the extension mode is periodization, each level of the decomposition will produce half of the size of the previous one. So if your original signal is of size 1024, after the first level decomposition you will get approximation and details of 512 samples, after the second level 256, after the third 128, and so on. But this is only if the extension mode is set to periodization. For other extension modes the output will be a little longer that half the previous one in each stage.
If you need the same number of coefficients in each scale, you should use an undecimated wavelet transform, also known as "algorithme a trous" or stationary wavelet transform, which is implemented in Matlab by the function swt().
In any case, I suggests you to read the theory for wavelet transform, as those are basic aspects that you ideally should know before trying to use a tool.
thank you so much sir.... I know that wavelet Decomposition breaks the signal like this that' s why I decided to opt for level 6 decomposition in order to get the delta frequency too, but the detail coefficients I am getting have very different dimensions so I just wat to sure ..that Are they OK
- Badges
- Science topic
- Similar topics
- Filing