Survey of Dirac: A Wavelet Based Video Codec for Multiparty Video Conferencing and Broadcasting - Intelligent Video Event Analysis and Understanding - page 221

Another view of wavelet transform is shown below:
Fig. 8 First level of an image decomposition by wavelet sub-band filters
The number of samples in each resulting subband is as implied by the diagram
above. The critical sub-sampling ensures that after each decomposition the
resulting bands have one quarter of the samples of the input signal.
5.2.1 Wavelet Filters
The choice of wavelet filters has an impact on compression performance, filters
having to have both compact impulse responses in order to reduce ringing artifacts
and other properties in order to represent smooth areas compactly. It also has an
impact on encoding and decoding speed in software. There are numerous filters
supported by Dirac to allow a trade-off between complexity and performance.
These are configurable in the reference software. These filters are all defined
using the 'lifting scheme' for speed [12].
5.2.2 Padding and Invertibility
Clearly, applying an N-level wavelet transform requires N levels of subsampling,
and so for reversibility, it is necessary that the dimensions of each component are
divisible by 2 N . So if this condition is not met in the size of input image/frame, the
input picture components are padded as they are read in, by edge values for best
compression performance.