| 摘要 |
In the hybrid encoder for video signals described the picture elements of a video picture are arranged into equally large data blocks and each newly entered data block, is compared with a data block of the previous video picture by means of a subtractor (SR). The data blocks of the previous picture are stored in a picture store (BS). In order to maintain the transmission bit rates as small as possible, the data blocks are conventionally subjected to a transformation (for example, a Fourier transformation) (F) and a subsequent quantization (Q). For the required strong bit rate reduction a rough quantization is necessary, which leads to a strong quantization noise. To eliminate the quantization noise a blockwise adaptive low-pass filtering (T,FTI) is performed in two signal paths of the hybrid encoder. The quantizer (Q) may then have a refined design or it may be omitted altogether. The result of the low-pass filtering (T,FTI) is a low-definition video picture which is, however, free from the occasionally occurring picture distortions caused by the quantization noise. |
