10.9 A Note on Signal Data Compression

Signal data compression is not among the main themes of this book; nevertheless, it is worth investigating briefly to see if there are any links with the analysis and restoration of signals. So far, we have dealt with signals having in mind either obtaining a 'better' signal (in a sense) or a signal description for analytical or diagnostic purposes. When any kind of processing or design was to be evaluated, the criteria of quality have been based on some mathematical concept. Primarily, some kind of absolute or mean-square error was used, based on the differences between the desired and obtainable results (restored signals, responses of designed systems, estimation of stochastic parameters etc.). These criteria are transparent and easy to implement; also, consistent theories can be built on them to enable formalised analysis or restoration. They are designed from this point of view and therefore they may look some-what arbitrary to a user of the signal.

The subjective quality of signals, as perceived by human beings via their sensory organs or as evaluated with respect to the practical use of the signal, need not necessarily correspond to the mathematically introduced criteria. The most prominent examples of this are multimedia signals acoustic signals and image sequences to be observed by human beings. As another example, the quality of digitally restored or decompressed medical signals, such as ECG or X-ray images as evaluated by medical staff, cannot be simply related to anything like the...