| 摘要 |
<p>A highly efficient coding/decoding by subjecting image signals to orthogonal transformation, by which all data of interlace images, inclusive of scarce and rich types of images in motions and even a mixed type thereof, can be coded efficiently. A circuit (19) senses motion vectors in every macro-block between an odd-numbered field and an even-numbered field. A circuit (21) determines the coding mode (coding of every field or coding of every frame), according to the median of the motion vectors. A control circuit (16) controls gates (15, 17) and switches (6, 9, 10, 15) corresponding to the determined coding mode, and it generates a predictive image of every field or frame from buffer memories (7, 8). Circuits from a difference circuit (22) to a VLC circuit (25) calculate the difference data between an image to be coded and the predictive picture thereof, subject the difference data to discrete cosine transformation, and thereafter perform variable-length coding. Also, the VLC circuit (25) sets the coding mode as a flag in the header of the predetermined hierarchy of a bit stream.</p> |
