| 摘要 |
In a predictive coding device responding to a sequence of image data signals representative of picture elements to produce a sequence of coded output signals, a plurality of provisional error signals are calculated for each picture element from the image data signals and different provisional prediction signals sent from predictors and are given weights in multipliers from a weight decision circuit in consideration of a position of each picture element to produce weighted error signals. The weighted error signals are summed up into a first sum signal by a first adder circuit and thereafter processed into the coded output signals and local decoded signals which are sent to the predictors. The predictors may carry out either different prediction from one another for each picture element or identical prediction in relation to different blocks. Alternatively, a motion vector which is calculated for each block may be smoothed in consideration of the other motion vectors calculated for adjacent blocks on motion-compensated prediction. Thus, weighted means values are calculated as the weighted error signals and/or smoothed motion vector at every picture element so as to mitigate a discontinuity among the blocks. |
