| 摘要 |
PURPOSE:To reduce the influence of an error and to easily decode a variable- length code by a method wherein an input signal is divided into small blocks they are rearranged and, when the size of the number of residual bits is smaller than a prescribed code length, the bits are separated once. CONSTITUTION:A small blocking part 2 changes a signal from an input part 1 into small blocks having the size of blocks for an orthogonal conversion; a rearrangement part 4 rearranges orthogonally converted components sequentially from a low component onward both horizontally and vertically toward a high component both horizontally and vertically; the components are output. Then, the orthogonal components are quantized at a quantization part 5; they are coded to variable-length codes at a variable-length coding part 6; at the same time, a code length is detected at a code-length detection part 7. In succession, a recording-block residual bit-number detection part 8 computes the residual number of bits on the basis of a code length which has been input. When the residual number of bits is larger than a maximum code length, a code word is written to a fixed-part memory 10 by using a SW 9. Inversely, when the residual number is small than the maximum code length, the code word is written to a separated-part memory 11. |
