A fixed-rate product pyramid vector quantization using a Bayesian model
Pavel Filip, Michael J. Ruf · 2003
The performance of the pyramid vector quantizer is analyzed. Some useful formulas for the computation of the pyramid numbers N(L, K) are derived from the generating function. The quantization error, when (L-K)-thresholding is employed is estimated. Extending the Laplacian model, the results can be generalized by applying the Bayesian method. Some simulation results for pyramid vector quantization in DCT (discrete cosine transform) image compression are given.>