Abstract: Blocks of each of successive frames of a motion picture signal are classified into classified blocks in accordance either with selected schemes for use in adaptive prediction encoding the signal or with whether the selected schemes are restricted or not. On encoding each classified block, a quantization parameter is decided based on an amount of codes produced per classified block. The codes are stored in a buffer, and delivered outwardly of the buffer at a predetermined code rate while the remaining codes are reserved in the buffer. The quantization parameter is also based on an initial amount of the reserved codes kept in the buffer at a beginning of each frame, a code amount characteristic value, and an amount of codes assigned to each classified block. The code amount characteristic value is calculated by a product of the quantization parameter and the amount of codes produced.

Abstract: In a moving picture coding device including a memory for storing a referenced picture, a motion vector search device comprises a primary predictive evaluation section and a subsidiary predictive evaluation section. Supplied with an input picture and connected to the memory, the primary predictive evaluation section carries out a primary motion vector search by generating primary predicted pictures using the input picture and the referenced picture to produce primary prediction error evaluated values. Connected to the primary predictive evaluation section, the subsidiary predictive evaluation section carries out a subsidiary motion vector search to estimate at least one subsidiary prediction error evaluated value using the primary prediction error evaluated values. The subsidiary predictive evaluation section therefore carries out the subsidiary motion vector search without generation of subsidiary predicted pictures.

Abstract: Blocks of each of successive frames of a motion picture signal are classified into classified blocks in accordance either with selected schemes for use in adaptive prediction encoding the signal or with whether the selected schemes are restricted or not. On encoding each classified block, a quantization parameter is decided based on an amount of codes produced per classified block. The codes are stored in a buffer, and delivered outwardly of the buffer at a predetermined code rate while the remaining codes are reserved in the buffer. The quantization parameter is also based on an initial amount of the reserved codes kept in the buffer at a beginning of each frame, a code amount characteristic value, and an amount of codes assigned to each classified block. The code amount characteristic value is calculated by a product of the quantization parameter and the amount of codes produced.