| 摘要 |
<p>PURPOSE:To attain data compression similar to the DPCM method while leaving some degree of the property of non-propagation of code error being the advantage of the PCM by predicting a coded value, and selecting the quantization characteristic in response to the predicted value. CONSTITUTION:A signal obtained from an encoder input terminal 4 is coded by a quantizer 5 selected adaptively by a switching signal being an output of a prediction device 6 among plural quantizers 5 and outputted from an encoder output terminal 7. The prediction device 6 outputs a prediction value predicting a signal obtained from the encoder input terminal 4 based on inter- frame correlation, inter-field correlation, inter-line correlation or inter-pixel correlation of a picture. The probability distribution of an input signal at the encoder input terminal 4 goes to high around the predicted value. Thus, the nonlinear quantizer quantizing the values around the predicted value in details is selected for the case to decrease the signal deterioration in data compression.</p> |
