An extrapolative-interpolative prediction coding method for HDTV signals

Techniques are introduced to solve two problems caused by the wide bandwidth of HDTV signals. One problem is that HDTV signals have more random noise than conventional TV signals, which causes coding efficiency degradation. The other is that very high sampling frequencies are required, which restricts the hardware implementation of the coding algorithm. To solve the first problem, it is proposed to use a spatio-temporal adaptive filter that can reduce random noise and improve coding efficiency. With regard to the second problem, an extrapolative and interpolative coding algorithm with adaptive intrafield/interframe prediction is presented. This algorithm eliminates the problem caused by the high sampling frequencies and provides efficient coding performance. A computer simulation is carried out to estimate the coding performance. Based on the simulation results, experimental coder hardware was implemented for 1125-line/60 Hz HDTV signals in order to evaluate real-time coding performance. The experimental results show that this coding method can provide good picture quality at around 90 or 135 Mb/s, and that HDTV digital transmission is very promising at the broadband ISDN H₄ rate