Inter-frame adaptive prediction system for television signals

Abstract

An inter-frame adaptive prediction system for television signals is disclosed in which a prediction value P.sub.1 for a picture element 1 in the current field is newly produced, through using prediction values P.sub.OPT and P.sub.MED obtained by the following two procedures. In a first procedure, the intra-field prediction (a prediction value X.sub.N), the inter-field prediction (a prediction value X.sub.K) and the inter-frame prediction (a prediction value X.sub.F) are performed for a picture element temporally ahead of the picture element in the current field F.sub.0, for instance, the picture element 2, and a prediction circuit which yields a prediction value closer to the picture element value X.sub.2 of the picture element 2 than the prediction values of any other prediction parts is selected, and a prediction value X.sub.1S for the picture element 1 obtainable from the selected prediction circuit is obtained and is used as the prediction value P.sub.OPT. In a second procedure, the intra-field prediction, the inter-field prediction and the inter-frame prediction are carried out for the picture element 1, and prediction values obtained from the three prediction circuits are compared in magnitude with one another to obtain a prediction value corresponding to a median, which value is used as the prediction value P.sub.MED. Therefore, for a picture signal such that the moving and stationary parts are clearly distinguished in the picture, P.sub.OPT obtained by the procedure 1 provides a prediction value close to the true value. On the other hand, for a picture signal such that the moving and stationary parts are not clearly distinguished in the picture, a weighed mean value of the predicted values P.sub.OPT and P.sub.MED is employed.