A video signal shuffling device forms a segment with blocks having a given offset value in horizontal and vertical directions from a given starting point uniquely determined on the screen, the offset value being identical in every region. The segment is formed in a unit of one frame or two frames. The shuffling device divides one frame comprising blocks into a plurality of regions. A coding device determines a final priority as well as a final quantizer thereby quantizing a segment and providing the segment with a variable-length-coding based on a first priority, coding-quantity obtained from a plurality of quantizers, second priority and a quantizer. A coding device and its decoder, the coding device carries out a hierarchical coding by only a motion vector in a compressed stream at a lower resolution side and by agreeing a motion detected and a DCT block size with a size on the screen.