Quantization

Abstract

We return to the model of continuous-tone full-motion pictures as a stochastic sequence of random variables, as described in Section 5.5. Pixel values are used as predictors, and thus:

$$ \chi _n {\mathbf{ }} = {\mathbf{ }}\chi _{n - 1} {\mathbf{ }} + {\mathbf{ }}\varepsilon _n $$

((6.1))

where:

• x_n is the pixel value to be predicted,

• x_n-1 is the pixel value to be used as predictor, and

• ɛ_n is the prediction error distribution.