Processing for compression - The MPEG - page 68

[ 5 ] Crochiere, R.E. and Rabiner, L.R., Interpolation and decimation of digital signals - a tutorial review. Proc. IEEE ,
69 , 300-331 (1981)
[ 6 ] Rabiner, L.R., Digital techniques for changing the sampling rate of a signal. In Digital Audio , edited by B. Blesser,
B. Locanthi and T.G. Stockham Jr, pp.79-89, New York: Audio Engineering Society (1982)
3.7 Downsampling filters
In compression pre-processing, it is often necessary to downsample the input picture. This may be needed to
obtain the desired picture size, e.g. 360 pixels across, from a standard definition picture which is 720 pixels across.
4:2:2 inputs will need vertical filtering to produce 4:2:0 which most of the MPEG-2 profiles use.
Figure 3.23 shows that when a filter is used for downsampling, the stopband rejection is important because poor
performance here results in aliasing. In an MPEG environment aliasing is particularly undesirable on an input signal
as after the DCT stage aliasing will result in spurious coefficients which will require a higher bit rate to convey. If
this bit rate is not available the picture quality will be impaired. Consequently the performance criteria for MPEG
downsampling filters is more stringent than for general use and filters will generally require more points than in
other applications.
Figure 3.23: Downsampling filters for MPEG prefiltering must have good stopband rejection to avoid allasing when
the sampling rate is reduced. Stopband performance is critical for a pre-processing filter.
3.8 The quadrature mirror filter
Audio compression often uses a process known as band splitting which splits up the audio spectrum into a series
of frequency ranges. Band splitting is complex and requires a lot of computation. One bandsplitting method which
is useful is quadrature mirror filtering. [ 7 ] The QMF is is a kind of twin FIR filter which converts a PCM sample stream
into two sample streams of half the input sampling rate, so that the output data rate equals the input data rate. The
frequencies in the lower half of the audio spectrum are carried in one sample stream, and the frequencies in the
upper half of the spectrum are carried in the other. Whilst the lower frequency output is a PCM band-limited
representation of the input waveform, the upper frequency output is not. A moment's thought will reveal that it could
not be because the sampling rate is not high enough. In fact the upper half of the input spectrum has been
heterodyned down to the same frequency band as the lower half by the clever use of aliasing. The waveform is
unrecognizable, but when heterodyned back to its correct place in the spectrum in an inverse step, the correct
waveform will result once more. Figure 3.24 shows how the idea works.