METHOD AND APPARATUS FOR COMPRESSING VIDEO SEQUENCES

摘要

A method for encoding a sequence of video images, each video image comprising pixels, said method comprising the steps of: representing the sequence by at least one set of consecutive images, each set of consecutive images comprising a plurality of video images; representing each set of consecutive images by one reference image and a subspace model of motion, the reference image and the subspace model of motion together comprising an IDLE model, wherein the subspace model of motion comprises a plurality of temporal coefficient vectors and a plurality of basis images, so that each element of each temporal coefficient vector corresponds to one video image and so that each element of each basis image contributes to the motion for a pixel, the combination of one temporal coefficient vector and one basis image together being called a factor, representing the sequence in encoded form by the collection of IDLE models.