PURPOSE: To obtain a sufficient prediction picture and to improve encoding efficiency by dividing a square block into two triangles and adaptively selecting a division line for the triangle, which generates the prediction picture by means of the shape conversion of the respective triangles, at every square block.
CONSTITUTION: A motion estimating part 1 estimates the motion of a grid point and the motion vectors of the respective grid points are obtained. A square block bi-seating part 3 divides the square block into the two triangles by a previously fixed method. An affine transformation deciding part 5 obtains the two parameter of affine transformation for one square block. An adaption selecting part 2 is the part adaptively selecting the division line of the square block at every square block. The square block bi-seating part 3 divides the square block in accordance with the selected division line with the division line selected by the adaption selecting part 2 and the motion vector as inputs.
KUSAO HIROSHI
AONO TOMOKO