General signal processing techniques that have been successfully applied to encode and compact audio information, such as Differential Pulse Code Modulation (DPCM), can be applied to the two-dimensional raster signal of each video frame. These usually use an average of the neighbouring pels as prediction and send only the prediction error. This mode, called intraframe prediction, exploits the spatial redundancy within each frame. If the predictor uses pels from different frames, usually the temporally previous to the current, it is called interframe prediction, and it removes temporal correlation as well.