Method for motion compensation of a moving image using two dimensional triangle patch wireframe model

A method for compensating a motion of a moving image after estimating said motion, comprising the steps of:

• (a) forming a triangle-patch wireframe model on a present frame of said moving image, wherein said triangle-patch wireframe model comprises apexes;
• (b) determining motion vectors between said apexes of said triangle-patch wireframe model and corresponding points on a previous frame of said moving image;
• (c) transforming said triangle-patch wireframe model on said present frame to form a transformed triangle-patch wireframe model on said previous frame by using said motion vectors such that said corresponding points respectively relate to secondary apexes of said transformed triangle-patch wireframe model;
• and
• (d) replacing a first pixel value of a first pixel located at a predetermined position within said triangle-patch wireframe model of said present frame with a replacement pixel value of a second pixel located at a corresponding position within said transformed triangle-patch wireframe model in said previous frame,
• wherein said apexes of said triangle-patch wireframe model comprise apex A, apex B, and apex C and position vectors of said apex A, said apex B, and said apex C respectively equal (x.sub.a,y.sub.a), (x.sub.b,y.sub.b), (x.sub.c,y.sub.c),
• wherein said predetermined position X of said first pixel has a position vector (x,y),
• wherein said secondary apexes of said transformed triangle-patch wireframe model comprise apex A', apex B', and apex C' and position vectors of said apex A', said apex B', and said apex C' respectively equal (x'.sub.a,y'.sub.a), (x'.sub.b,y'.sub.b), (x'.sub.c,y'.sub.c),
• wherein said corresponding position X' of said second pixel has a position vector (x',y'), and
• wherein a motion vector between said predetermined position X and said corresponding position X' equals
• (1-p-q)(x.sub.a -x.sub.a ',y.sub.a -y.sub.a ')+p(x.sub.b -x.sub.b ',y.sub.b -y.sub.b ')+q(x.sub.c -x.sub.c ',y.sub.c -y.sub.c '),
• wherein p equals ((x-x.sub.a)v.sub.y1 -(y-y.sub.a)v.sub.x1)/(v.sub.x0 v.sub.x1 -v.sub.x1 v.sub.y0), q equals ((y-y.sub.a)v.sub.x0 -(x-x.sub.a)v.sub.y0)/(v.sub.x0 v.sub.y1 -v.sub.x1 v.sub.y0), v.sub.x0 equals x.sub.b -x.sub.a, v.sub.y0 equals y.sub.b -y.sub.a, v.sub.x1 equals x.sub.c -x.sub.a, and v.sub.y1 equals y.sub.c -y.sub.a, and
• wherein 0.ltoreq.p.ltoreq.1, 0.ltoreq.q.ltoreq.1, and 0.ltoreq.p+q.ltoreq.1.