High level decomposition in dualtree2

I am trying to do complex wavelet transform without down sampling at every level. In the matlab code dualtree2, the first and higher level decomposition uses different filter functions. For the higher level decomposition, the LL subband goes through a polyphase filtering which down samples the input before down sampling again with quad2complex. What is the purpose of this polyphase filtering and is it possible to do the polyphase filtering without down sampling?

Answer 1

[Vatsal]

Hi,

I understand that you want to know the purpose of polyphase filtering in higher-level decomposition and whether it is possible to perform it without downsampling.

In the "dualtree2" function, polyphase filtering is used to divide the input signal into multiple substreams, each containing a subset of the original samples. These substreams are then individually filtered using different filter functions. This approach allows for parallelization of the filtering operations, resulting in faster computation.

While it is indeed possible to perform polyphase filtering without downsampling, downsampling is often employed to reduce the computational load. Downsampling reduces the number of samples that need to be processed in subsequent stages, making the algorithm more efficient in terms of computational resources.

To learn more about “dualtree2” usage and syntax, you may refer to the MathWorks documentation link below: -

https://in.mathworks.com/help/wavelet/ref/dualtree2.html

I hope this helps!