Adaptive Prediction for High-Quality Television Transmission Coding Based on the LMS Algorithm

Differential pulse-code modulation (DPCM) and hybrid transform coding are two popular high-performance methods of reducing the bit rate of component or composite digital television signals for point-to-point transmission. Both methods involve a predictor and subsequent coding of the resulting prediction error signal. This article describes a new prediction strategy, based on the least-mean-square (LMS) algorithm, which optimizes the accuracy of a multielement linear predictor by allowing it to adapt continuously to local picture-signal statistics. Performance benefits of using the algorithm in both coding methods are evaluated, based on results of computer simulations. The inherent problems of decoder instability and methods of overcoming them, along with possible hardware configurations of practical implementations, are described.

Print ISSN

0036-1682

Published

1989-08

Content type

Original Research

DOI

10.5594/J02692