The way to reduce the memory used in digital video codec is disclosed. The method generally (A) ( ) comprises a two-dimensional grid ( ) vertical down sampling performed by the sampling and a second down a first reference picture using a horizontal -down can not be obtained by performing a sampling pattern otherwise generating a reference image ; (B) generating a third reference image by sampling up the second reference image; And (C) is characterized in that using the third reference picture , comprising the step of processing the image to a video signal .
展开▼