| 摘要 |
PROBLEM TO BE SOLVED: To more exactly describe the movement of an object and to improve the quality of a predicted image by using a motion vector field pattern as additional information in addition to a parallel moving. SOLUTION: In a coder, a coding object image 1 is inputted with a motion vector field pattern 2 in a motion detection part 3 at first. For every block, a motion vector and a motion vector field pattern code 4 are determined. The determined motion vector and the motion vector field code 4 are inputted in a picture element unit motion vector generator 5, the motion vector 6 of a picture element unit is determined again, a movement compensation is performed in a picture element unit movement compensation part 7 and a predicted image 8 is determined. On the side of a decoder, the motion vector, the motion vector field pattern code 4 and the movement compensation predictive error 10 determined from the coding object image 1 and a predicted image 8 by a subtracter 9 are multiplexed and transmitted. Thus, movements such as the rotation/ elongation and contraction/magnification of an object can be efficiently predicted. |
