Method of coding and decoding image
View Patent ↗A method of simplifying the arithmetic operation in a global motion compensation process approximates the motion vector field of the whole image without using many parameters. Motion vectors in the global motion compensation are found by the interpolation and/or extrapolation of the motion vectors of a plurality of representative points 602, 603 and 604 having particular features in the spatial distance thereof. Since the shift operation can be substituted for the division for synthesizing a predicted image of global motion compensation, the processing using a computer or a dedicated hardware is simplified.
1. A coding method comprising:
storing a decoded image of a previously decoded frame as a reference image,
synthesizing a predicted image of a current frame by performing motion estimation between an input image of the current frame and the reference image, wherein the synthesizing step includes:
calculating motion vectors of representative points having coordinates (i, j), (i+p, j), and (i, j+q), wherein i and j are integers, and p and q are powers of integers of 2 by using motion vectors of corner points of the predicted image of the current frame;
calculating a motion vector of each pixel in the predicted image of the current frame by using the motion vectors of representative points; and
synthesizing the predicted image by using the motion vector of each pixel and the reference image.
2. A coding method according to claim 1 , wherein:
the horizontal and vertical components of said motion vector of each pixel are integral multiples of 1/m;
the horizontal and vertical components of said motion vectors of representative points are integral multiples of 1/k; and
said motion vector of each pixel is calculated according to equations:
u ( x,y )=(( u 0 pq +( u 1 −u 0)( x−i ) q +( u 2 −u 0)( y−j ) p ) m )//( pqk ),
v ( x,y )=(( v 0 pq +( v 1 −v 0)( x−i ) q +( v 2 −v 0)( y−j ) p ) m )//( pqk ),
wherein (u, (x, y), v(x, y)) is m times of a motion vector of a pixel having coordinates (x, y), (u0, v0), (u1, v1), and (u2, v2) are k times of said motion vectors of said representative points (i, j), (i+p, j), and (i, j+q) respectively, m and k are powers of integers of 2, “//” represents integer division that rounds the quotient of ordinary division into an adjacent integer when the quotient of ordinary division is not an integer, and its priority as an operator is the same as that of ordinary multiplication and division, and u (x,y), v(x,y), u0, v0, u1, v1, u2 and v2 are integers.