Adaptive Intra-Interframe DPCM Coder

The bit rate required for digital transmission of television pictures can be significantly reduced by interframe DPCM encoding. The coding method which has been widely proposed for video-telephone and video-conference application is conditional replenishment.1,2 In conditional replenishment, each frame of a television sequence is segmented into changed and unchanged areas. Various methods can be used for encoding the changed parts of a frame. Intraframe predictive coding is very efficient for these parts. M In conditional replenishment, no information about the unchanged areas is transmitted. At the receiver, the unchanged areas are reconstructed by repeating from the previous frame. However, it is necessary to transmit address information that 747 indicates the location of the changed areas. Several modifications and improvements of the basic method of conditional replenishment have been made. Most by them are described in a survey by Haskell.5 This paper describes adaptive intra-interframe prediction. It is obvious that stationary background of a frame is best predicted from a pel in the previous frame which has the same position as the pel to be predicted, whereas parts of a frame with moving objects are better predicted by an intraframe predictor. Therefore, a prediction scheme which provides automatic switching between the two types of predictors, depending upon the part of the picture, will result in lower bit rates. To avoid the transmission of additional predictor control information, the adaptive prediction schemes described here are based on previously transmitted reconstructed pels.