Error using newly added wavelet to wavemngr

I am working with a tool in matlab that matches an orthogonal wavelet to a given prototype signal. The attatched image depicts the information stored within the waveletFilter structure generated by the tool.

I would like to add the f/g filters to wavemngr(). I begin by saving f and calling the .mat file by the wavelet short name I plan to use. I then run the following script:

wsname = 'L1F';

wname = 'L1OptimizedFilter';

eval([wsname '=waveletFilterL1.f/sqrt(2);']);

save(wsname, wsname);

wavemngr('add', wname, wsname, 1, '', [wsname '.mat']);

When I use wavemngr('read') to validate the addition of my filter, it is there.

However, when I then proceed to access any information regarding the properties, or use the filter with a given transform, I get an error message saying my filter has not been found.

[Umar]

Hi Dejan,

Have you tried directly saving the filter data to a .mat file and then loading it into wavemngr() without dynamic variable creation, the reason I am asking this question is because if you are using eval() in your code, it dynamically creates the variable name which could not be correctly linking it to the tool's internal structure, leading to the tool's inability to recognize the added filter. Here is an updated script,

wsname = 'L1F';
wname = 'L1OptimizedFilter';
filterData = waveletFilterL1.f / sqrt(2);
save([wsname '.mat'], 'filterData');
wavemngr('add', wname, wsname, 1, '', [wsname '.mat']);

Hope this should help you resolve your problem. Please let me know if you have any further questions.

[Dejan Vlasakov]

Hello, thank you for the response!

Sadly the same error persists. I would like to clarify, the tool generates the two filters. The problem appears when I have saved them to wavemngr, and then try to use them for decompositon with wpdec().

[Walter Roberson]

Please read http://www.mathworks.com/matlabcentral/answers/304528-tutorial-why-variables-should-not-be-named-dynamically-eval for information about why we strongly recommend against creating variable names dynamically.

[Umar]

Hi Dejan,

There seems to be discrepancies in your statements, You mentioned, I would like to add the f/g filters to wavemngr(). I begin by saving f and calling the .mat file by the wavelet short name I plan to use. I then run the following script: wsname = 'L1F'; wname = 'L1OptimizedFilter'; eval([wsname '=waveletFilterL1.f/sqrt(2);']); save(wsname, wsname); wavemngr('add', wname, wsname, 1, '', [wsname '.mat']); So, could you share the script code for wname = 'L4F’

Then, you mentioned, “when I then proceed to access any information regarding the properties, or use the filter with a given transform, I get an error message saying my filter has not been found.” But the screenshot that you have shared for error message contradicts with your statement,it is telling you that you are using invalid wavelet name, it sounds like either you ran the code earlier using invalid wavelet name and sharing this new line of code tobj=wpdec(s,5,”L40optimizedFilter”); because this does not making any sense saying and you are declaring “L4OptimizedFilter” but that does not mean filter not found. Error using wavemngr,invalid wavelet name “l40optimizedfFilter” So, there is mismatch in problem description mentioned by you. Now, your recent comments mentioned that you are facing problems with both filters but not clear about if errors are the same. So, my suggestion would be edit your comments sharing the right filter name with errors, sharing the code file which is causing the error. The more details you will provide will help me troubleshoot your problem better.