Method for motion-compensated frame-to-frame prediction coding

Claim 2

In a method of motion-compensated frame-to-frame prediction coding, wherein prediction errors are coded by and transmitted from a channel coder and the prediction errors are calculated as a function of the signal content of a current frame and a prior frame with a predictor element, the method including breaking down the entire current frame into a regular block scanning field of blocks of n×n pixels and ascertaining by interval calculation directional displacements for each block relative to respective blocks of a previous frame, the improvement comprising the steps of: combining adjoining blocks for which the directional displacements ascertained by interval calculation are not equal to zero into respective complexes of blocks, calculating by regression calculation parameters of a motion principle of one of the complexes from the respective directional displacements ascertained by interval calculation of the blocks which belong to the respective one of the complexes, calculating an estimated frame with the aid of individual block displacements derived from the regression calculation parameters of the motion principle of the one of the complexes, and controlling the predictor element with the aid of the estimated frame calculation.

Claim 3

A method as in claim 1, wherein said step of calculating the estimated frame includes the steps of ascertaining by interval calculation for each block whether the ascertained directional displacement for each block or the block displacements derived by regressive calculation of the parameters of the motion principle produces a smaller prediction error in the location of the block relative to the actual location of a corresponding block in the current frame, ascertaining whether no displacement or a nullification of a prediction of the location of the block in the estimated frame produces a smaller prediction error, and for the calculation of each block of the estimated frame, ascertaining and using the one of the ascertained directional displacement for each block, the block displacement derived by regressive calculation of the parameters of the motion principle, no displacement or a nullification of a prediction of the location of the block in the block in the estimated frame, which produces the smallest prediction error.

Claim 4

A method as in claim 1, further comprising the step of performing a correct-motion interpolation of possibly uncoded and untransmitted partial and full frames using selected and transferred displacement vectors from a decoder to which the signal is being sent.

Claim 5

In a method of motion-compensated frame-to-frame prediction coding, wherein prediction errors are coded by and transmitted from a channel coder and the prediction errors are calculated as a function of the signal content of a current frame and a prior frame with a predictor element, the method including breaking down the entire current frame into a regular block scanning field of blocks of n×n pixels and ascertaining by interval calculation directional displacement for each block relative to respective blocks of a previous frame, the improvement comprising the steps of: combining adjoining blocks for which the directional displacements ascertained by interval calculation are not equal to zero into respective complexes of blocks, calculating by regression calculation parameters of a motion principle of one of the complexes from the respective directional displacements ascertained by interval calculation of the blocks which belong to the respective one of the complexes, predicting each block of a first predicted frame using block-by-block motion compensation, as a function of the directed displacements, predicting each block of a second predicted frame using object-related motion compensation, using a displacement vector calculated in accordance with the parameters calculated for the motion principle derived for one of the complexes, for each block of the first and second predicted frames, comparing the block to the corresponding block of the current frame to determine a respective prediction error in the location thereof, selecting one of block-by-block motion compensation and object-related motion compensation depending on the relative size of the prediction errors associated with the corresponding blocks of the first and second predicted frames, and applying a signal corresponding the selected motion compensation to the predictor.

Claim 6

In a method of motion-compensated frame-to-frame prediction coding, wherein prediction errors are coded by and transmitted from a channel coder and the prediction errors are calculated as a function of the signal content of a current frame and a prior frame with a predictor element, the method including breaking down the entire current frame into a regular block scanning field of blocks of n×n pixels and ascertaining by interval calculation directional displacement for each block relative to respective blocks of a previous frame, the improvement comprising the steps of: combining adjoining blocks for which the directional displacements ascertained by interval calculation are not equal to zero into respective complexes of blocks, calculating by regression calculation parameters of a motion principle of one of the complexes from the respective directional displacements ascertained by interval calculation of the blocks which belong to the respective one of the complexes, predicting each block of a first predicted frame using block-by-block motion compensation, as a function of the directed displacements, predicting each block of a second predicted frame using object-related motion compensation, using a displacement vector calculated in accordance with the parameters calculated for the motion principle derived for one of the complexes, for each block of the first and second predicted frames, comparing the block to the corresponding block of the current frame to determine a respective prediction error in the location thereof, selecting one of block-by-block motion compensation and object-related motion compensation depending on the relative size of the prediction errors associated with the corresponding blocks of the first and second predicted frames, and applying a signal corresponding the selected motion compensation to the predictor.

Claim 7

A method as in claim 4, further comprising the step of performing a correct-motion interpolation of possibly uncoded and untransmitted partial and full frames using selected and transferred displacement vectors from a decoder to which the signal is being sent.

Claim 8

A method as in claim 4, wherein the steps of comparing and selecting include the steps, for each block, of: (a) ascertaining by interval calculation whether object-related motion compensation or block-by-block motion compensation produces the smaller prediction error and (b) selecting the motion compensation which produces the smaller prediction error.

Claim 9

A method as in claim 6, further comprising the step of weighting the selection of motion compensation in favor of object-related motion compensation.

Claim 10

A method as in claim 6, wherein the step of calculating the parameters of the motion principle comprises the step of calculating the parameters of the motion principle solely from ascertained displacements of corners of the contents of the current frame.

Claim 11

A method as in claim 6, further comprising the step of performing a correct-motion interpolation of possibly uncoded and untransmitted partial and full frames using selected and transferred displacement vectors from a decoder to which the signal is being sent.

Claim 12

A method as in claim 4, wherein the steps of comparing and selecting include the steps, for each block, of: (a) ascertaining by interval calculation whether object-related motion compensation or block-by-block motion compensation produces the smaller prediction error, (b) ascertaining whether prediction error produced by no displacement and by a nullification of the prediction leads to smaller prediction errors, and (c) selecting for motion compensation the one of object-related motion compensation, block-by-block motion compensation, no displacement and nullification of the prediction which produces the smallest prediction error.

Claim 13

A method as in claim 10, further comprising the step of weighting the selection of motion compensation in favor of object-related motion compensation.

Claim 14

A method as in claim 10, wherein the step of calculating the parameters of the motion principle comprises the step of calculating the parameters of the motion principle comprises the step of calculating the parameters of the motion principle solely from ascertained displacements of corners of the contents of the current frame.

Claim 15

A method as in claim 10, further comprising the step of performing a correct-motion interpolation of possibly uncoded and untransmitted partial and full frames using selected and transferred displacement vectors from a decoder to which the signal is being sent.