| 摘要 |
<p>PROBLEM TO BE SOLVED: To improve prediction precision and to improve performance for correcting an error when a classification adaptive prediction processing is adapted to a digital information signal after an encoding/decoding processing. SOLUTION: A decoded picture signal and additional information for decoding are outputted from a decoder 1. Additional information is signal type information, picture format information, picture quality information and a movement vector, etc. A class based on the additional information is generated. Featured values are extracted from pixel data of a class tap which is extracted by an area segmenting part 4. A class code is generated based on the additional information class, the featured values and an error flag indicating the position of an error correction impossible error pixel. A prediction coefficient ROM 8 outputs a prediction coefficient set corresponding to the supplied class code to a prediction arithmetic part 9. The prediction coefficient is previously decided by learning and stored. Predicting calculation is performed by pixel data of a prediction tap which is extracted by a prediction tap data generating part 5 and by the prediction coefficient set supplied from the ROM 8 so that the picture signal where the error is corrected is generated concerning the output picture signal of the decoder 1.</p> |
