摘要 |
The method involves selecting one of two images (n,n-k) as a reference for a complete triangulation operation from a set of points chosen according to predetermined characteristics. The peaks are subjected to motion estimation for an intermediate image (n-k+p) to generate displacement vector estimates from which the co-ordinates of the displaced peaks are calculated. The vectors are then calculated for each vertex of the triangles of the image to be interpolated, with the deformation of each triangle represented by a transformation which is suitable for rotations, translations and zooms. |