Method for transformation-coding full motion image sequences
AssigneeRobert Bosch GmbH
InventorsAlexander Romanowski, Sven Bauer, Peter Siepen, Mathias Wien, Thomas Wedi
Life of this patent
Estimating the term…
Abstract
In the case of a method for transformation encoding of moving-image sequences, motion vectors are estimated block-by-block, with which said motion vectors a motion compensation is carried out. The prediction error is transformation-encoded. According to the invention, the block size of the transformation encoding is coupled to the block size used in each case for the motion compensation. This measure makes it possible to enhance the efficiency of the encoding of the prediction error in hybrid encoding procedures that use various block sizes.
Claims8
A method for transformation encoding of moving-image sequences, comprising the following steps:
- estimating motion vectors block-by-block between a reference image signal and an actual image signal from the image sequence, with which said motion vector compensation is carried out, wherein individual frames of the moving image sequence are divided into macroblocks, whereby for each macroblock the block size of a macroblock, square sub-blocks of a macroblock, or rectangle sub-blocks of a macroblock are adaptively used for motion compensation, and
- transformation-encoding the residual prediction error with the following measure:
- the block size of the transformation encoding is coupled to the block size used in each case for the motion compensation, whereby a maximum portion of the prediction error is commonly transformed, wherein the block transformation is selected for each macroblock that has the same block size as the block size used for the motion compensation.
Dependencies are inferred from the claim text.