Limiting error based optimisation of transformation algorithms for measurement signal compression

The parameter optimisation problems of lossy transformation compression algorithms for measurement signals is presented in the paper. The notional parameters should be understood as operational variables such as. e.g., transformation type, transformation order of P, or a type of the applied quantization procedure of transform coefficients. A selection criterion of parameters should include not only the obtained compression ratio and the losses in the signal undergoing a compression decompression procedure, but also the type of test signal and the error measure formula. Having that in mind, an error limiting criterion and testing signals of synthetic character were proposed for optimising the compression algorithms. The application of such algorithms to test the lossy compression algorithms makes the results obtained in such a way independent of the shape of a single signal and makes them valid for the whole class of measurement signals while the determined limiting errors assume the character of maximum errors. By varying the parameters of the synthetic signals the operation of algorithms can be tested within certain intervals of signal parameter variability, not for single values of these parameters but as it occurs for real signals.