| 摘要 |
PROBLEM TO BE SOLVED: To reduce the amount of calculation and to execute an inexpensive software processing by dividing an input picture, generating a block, executing an inter-frame prediction processing, a DCT processing and a quantization processing in a block unit, enlarging an obtained quantization coefficient in a frequency space and executing a variable length encoding processing. SOLUTION: A blocking means 201 immediately divides an input picture and generates a block constituted of M-pieces of horizontal pixels and N-pieces of vertical lines. An orthogonal conversion coefficient block is generated in a DCT(discrete cosine transformation) 203 in the block unit. A quantization means 204 generates a quantization coefficient block. In the quantization coefficient block, a reference picture is generated in an inverse quantization means 205 and an IDCT(inverse discrete cosine transformation) 206, and a moving vector is calculated in an inter-frame prediction means 202 by using an updated reference picture 207. The quantization coefficient block is doubled in the frequency space by a vertical double-enlarging processing 208, the obtained quantization coefficient block is variable length-encoded 209 and compressed picture data is generated. |
