FOERFARANDE FOER SOEKNING AV ROERELSEVEKTORER FOER EN VIDEOSIGNAL.

摘要

The invention relates to a method for the search of a motion vector (MV) for a video signal in hybrid coding where the temporal redundancy of the image is removed by DPCM prediction and a resulting prediction error is converted by a transform operation into transform-domain coefficients and transmitted to a decoder in the form of coefficients. To avoid the problems caused in hybrid coding by motion vectors found by the commonly used block matching method, a first operation approximating the transform of the actual hybrid encoder is applied (35) to a difference image produced by an image block corresponding to a motion vector candidate to be tested in the search of the motion vectors, and the coefficients produced by said operation are used to approximate a number of bits required to code the transform coefficients, and a block minimizing said number of bits is selected from within a selected motion vector search area as the prediction block. <IMAGE>