Quantization of random sequences and related statistical problems
Oleg Seleznjev, Mykola Shykula · 2006
We consider quantization of signals in probabilistic framework. In prac-tice, signals (or random processes) are observed at sampling points. We study probabilistic models for run-length encoding (RLE) method. This method is characterized by the compression efficiency coefficient (or quan-tization rate) and is widely used, for example, in digital signal and image compression. Some properties of RLE quantization rate are investigated. Statistical inference for mean RLE quantization rate is considered. In particular, the asymptotical normality of mean RLE quantization rate es-timates is studied. Numerical experiments demonstrating the rate of con-vergence in the obtained asymptotical results are presented.