Wavelet toolbox waveletAnalyzer app vs wdencmp signal decomposition length not the same

1 次查看（过去 30 天）

Hadi Jaber 2018-10-13

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/423807-wavelet-toolbox-waveletanalyzer-app-vs-wdencmp-signal-decomposition-length-not-the-same

编辑： Hadi Jaber 2018-10-13

tldr; the wavelet app has same size detail coefficients while wdencmp() has sampled halved per filter bank as one would expect. is the app using a different algorithm than the wdencmp() DWT to deconstruct signal?

Hello, I am trying to understand the wavelet toolbox better however there is something that I cant seem to figure out. In the waveletAnalyer I analyzed a signal using a db4 wavelet with 5 level filter bank

(image.png) Now I did the same thing in command line with assumption that this was the same algorithm the app uses. here is my code

in the code you can see that I used dwt() for the transformation, this is the same functino that wdencmp() calls to destruct the signal. here are the plots of the detail coefficients using the wdencmp() function

My question is: why are the detail coefficients of different lengths from app vs function. You would expect that the detail coefficients should be sampled by half for every level however in the app the size of all detail coefficients are same while in function the sizes are sampled by two as you would expect from a DWT algorithm. Is the app just not using the same algorithm? If not then how what is it using? my best guess is that it is using a CWT as I am still new to wavelets. I appreciate any help.