Other Techniques and Current Research

Apart from transform coding, other video coding techniques include block-based and non-block-based algorithms. An example of the former is fractal encoding, which segments the image in small enough blocks to apply self-similarity theory, and finally describe the whole image with a formula. This leads to very high compression efficiency, but it is also a very computationally intensive task at the encoder. Subband coding is an example of the latter, which applies several bandpass filters to the source image, producing bands of spatial frequencies, which are further compressed in the same manner as the DCT coefficients.

All the techniques described above are content-independent, that is, they deal with any type of source images, as they apply generic waveform encoding. For special applications as personal communications, model-based coding schemes are applied, which achieve very high compression efficiency, leading to very low bitrate coding schemes. For example a human speaker is usually modelled as `head and shoulders before a stationary background' video sequences, and geometrical transforms are applied to handle temporal changes, as lips and eyes movement.

Hosted by www.Geocities.ws