| 摘要 |
A video compression method, which can be applied especially in the transfer of a video signal at low transfer rates. In the method, the video picture is divided into blocks, and in order to code the picture information of a block, the pattern information of the block and the value information connected with it are first derived from the picture information of the block. For the pattern defined by the pattern information, the pattern model, which most resembles it and for which a corresponding short code has been defined, is searched for from a predefined small group of block pattern models. The code of the most resembling pattern model is set as the pattern information code, and the value information code corresponding with the value information is combined with it. In connection with the method, different methods for a closer definition can be used, if the difference between the block pattern and the block pattern model exceeds a limiting value that has been set. In reconstructing the block, the corresponding pattern model is searched for, and it is combined with the value information corresponding with the value information code. The value information of the pixels of the block and the location of the contours can be more closely defined during reconstruction by interpolating on the basis of the value information of the adjacent blocks. |
