perform_quantization.m

来自「基于wavelet和bandelet的spiht图像压缩！」· M 代码 · 共 18 行

function [y, nbr_bits, nbr_bits_tot, y_quant] = perform_quantization(x, T)

% perform_quantization - perform a nearly uniform quantization of the signal.
%
%   [y, nbr_bits, nbr_bits_tot, y_quant] = perform_quantization(x, T);
%
%   The quantizer is defined by y=Q_T(x) where:
%       Q_T(x) = 0    if  |x|<T
%       Q_T(x) = sign(x) * ([x/T]+0.5)*T      where [.]=floor
%   (i.e. a nearly uniform quantizer with twice larger zero bin).
%
%   y is the quantified value
%   nbr_bit as same size as x and store the number of bits needed to code
%       each entry of x
%   nbr_bits_tot is just sum(nbr_bits_tot(:))
%   y_quant is the signed token representing each entry of y.
%
%   Copyright (c) 2004 Gabriel Peyr