Method and apparatus for video signal encoding with motion compensation

Claim 2

Apparatus for encoding a video signal including a plurality of samples each representing the intensity of an element of a picture, including: means (220) for estimating the displacement of objects in said picture which occur between the present frame and a previous frame; first means (210, 226) for predicting the intensities of a set of previously processed ones of said pels based upon the intensities at displaced locations in said previous frame determined by said estimates; second means (210, 213) for predicting the intensities of said first set of pels based upon the intensities of the spatially corresponding locations in said previous frame; means (250) for computing first and second sums of the prediction errors associated with predictions made by said first and second means using displaced frame difference and frame difference techniques, respectively; and means (240) responsive to said computing means for predicting the intensity value of the present pel using the one of said prediction techniques which has the smaller prediction error sum.

Claim 3

The invention defined in claim 1 wherein said apparatus further includes: means (202) for forming the prediction error between the intensity value of said present pel and the predicted value thereof; and means (206, 208) for encoding said prediction error only if it exceeds a threshold value.

Claim 4

The invention defined in claim 2 wherein said apparatus further includes means (408) for encoding the locations of pels for which said prediction error exceeds said threshold.

Claim 5

The invention defined in claim 3 wherein said displacement estimating means includes: means (221) for storing a previous estimate of displacement; and means (223) responsive to said previous estimate for recursively updating said estimate.

Claim 6

Apparatus for encoding a series of samples each representing the intensity of an element of a picture, comprising: means for predicting the intensity values of a set of previously encoded pels using first and second predictors; means for selecting as the predicted value of the presently processed pel the value derived from the one of said first and second predictors for which the sum of prediction errors for said previously encoded set of pels is smallest; and means for encoding (a) the prediction error of unpredictable pels for which the prediction error for said presently processed pel exceeds a threshold, and (b) the locations of said unpredictable pels.

Claim 7

Apparatus for encoding a series of samples each representing the intensity of an element of a picture, comprising: means for predicting the intensity values of a set of previously encoded pels using first and second predictors; means for selecting as the predicted value of the presently processed pel the value derived from the one of said first and second predictors for which the sum of prediction errors for said previously encoded set of pels is smallest; and means for encoding (a) the prediction error of unpredictable pels for which the prediction error for said presently processed pel exceeds a threshold, and (b) the locations of said unpredictable pels.

Claim 8

The invention defined in claim 5 wherein said apparatus further includes means for estimating the displacement of objects in said picture, and wherein: said first predictor is based upon the intensity of locations in a previous frame of said picture which spatially correspond to the locations of said set of previously encoded pels; and said second predictor is based upon the intensity of locations in said previous frame displaced from said spatially corresponding locations by an amount determined by said displacement estimating means.

Claim 9

The invention defined in claim 6 wherein said prediction error encoding means includes a variable length encoder and said location encoding means includes a run length encoder.

Claim 10

The invention defined in claim 6 wherein said displacement estimating means is arranged to recursively update each estimate based upon the value of the prior estimate.

Claim 11

Apparatus for encoding a video signal which includes a plurality of samples each representing the intensity of a particular element of a picture comprising: means (220) for forming an estimate of the displacement of objects in said picture between a present frame and a prior frame; means (226, 210, 213, 211) for forming first and second predictions of the intensities of a series of pels in said present frame based upon (a) corresponding locations in said prior frame, and (b) displaced locations in said previous frame determined by said displacement estimates: means (250) for selecting a predicted value for the present pel based upon the relative magnitudes of the prediction errors of said first and second prediction means for comparing the prediction error associated with said selected predicted value to a threshold; and means (206) for encoding the location and prediction error of pels for which said prediction error exceeds said threshold.

Claim 12

Apparatus for encoding a video signal which includes a plurality of samples each representing the intensity of a particular element of a picture comprising: means (220) for forming an estimate of the displacement of objects in said picture between a present frame and a prior frame; means (226, 210, 213, 211) for forming first and second predictions of the intensities of a series of pels in said present frame based upon (a) corresponding locations in said prior frame, and (b) displaced locations in said previous frame determined by said displacement estimates: means (250) for selecting a predicted value for the present pel based upon the relative magnitudes of the prediction errors of said first and second prediction means for comparing the prediction error associated with said selected predicted value to a threshold; and means (206) for encoding the location and prediction error of pels for which said prediction error exceeds said threshold.

Claim 13

The invention defined in claim 9 wherein said encoding means includes a variable length encoder for encoding said prediction error information and a run length encoder for encoding said location information.

Claim 14

The invention defined in claim 9 wherein said displacement estimation means includes means for updating the prior displacement estimate in moving areas of said picture.

Claim 17

The method defined in claim 12 wherein said predicting step further includes the steps of: estimating the displacement of objects in said picture; p1 forming a first prediction based upon the intensity of locations in a previous frame of said picture which spatially correspond to the locations of said set of previously encoded pels; and forming a second prediction based upon the intensity of locations in said previous frame displaced from said spatially corresponding locations by an amount determined in said displacement estimating step.

Claim 18

The method defined in claim 13 wherein said encoding step includes variable length encoding said prediction error values and run length encoding said locations.

Claim 19

The method defined in claim 13 wherein said estimating step includes recursively updating each estimate based upon the value of the prior estimate.

Claim 20

Apparatus for encoding a video signal including a plurality of samples each representing the intensity of an element of a picture, including: means (220) for estimating the displacement D of objects in said picture which occur between the present frame at time t and a previous frame at time t-τ; first means (226) for predicting the intensities I(x A ,t), I(x B ,t), I(x C ,t) . . . of a set A, B, C . . . of said pels at locations x A , x B , x C . . . based upon the intensities I(x A -D,t-τ), I(x B -D,t-τ), I(x C -D,t-τ) . . . at displaced locations x A -D, x B -D, x C -D . . . in said previous frame determined by said estimates; second means (210, 213) for predicting the intensities of said first set of pels based upon the intensities I(x A ,t-τ), I(x B ,t-τ), I(x C ,t-τ) . . . of the spatially corresponding locations x A , x B , x C . . . in said previous frame; means (250) for computing first and second sums of the prediction errors associated with predictions made by said first and second means, respectively; and means (240, 206) responsive to said computing means for encoding the error value ε representing the difference between the present sample of said signal which represents the intensity value I(x Z ,t) at location x Z and a predicted version P Z thereof, wherein P Z is predicted based upon the intensity I(x Z ,t-τ) of the spatially corresponding location x Z in the previous frame or the intensity I(x Z -D,t-τ) of the displaced location x Z -D in said previous frame, depending upon which of said first and second sums is smaller.

Claim 21

Apparatus for encoding a video signal including a plurality of samples each representing the intensity of an element of a picture, including: means (220) for estimating the displacement D of objects in said picture which occur between the present frame at time t and a previous frame at time t-τ; first means (226) for predicting the intensities I(x A ,t), I(x B ,t), I(x C ,t) . . . of a set A, B, C . . . of said pels at locations x A , x B , x C . . . based upon the intensities I(x A -D,t-τ), I(x B -D,t-τ), I(x C -D,t-τ) . . . at displaced locations x A -D, x B -D, x C -D . . . in said previous frame determined by said estimates; second means (210, 213) for predicting the intensities of said first set of pels based upon the intensities I(x A ,t-τ), I(x B ,t-τ), I(x C ,t-τ) . . . of the spatially corresponding locations x A , x B , x C . . . in said previous frame; means (250) for computing first and second sums of the prediction errors associated with predictions made by said first and second means, respectively; and means (240, 206) responsive to said computing means for encoding the error value ε representing the difference between the present sample of said signal which represents the intensity value I(x Z ,t) at location x Z and a predicted version P Z thereof, wherein P Z is predicted based upon the intensity I(x Z ,t-τ) of the spatially corresponding location x Z in the previous frame or the intensity I(x Z -D,t-τ) of the displaced location x Z -D in said previous frame, depending upon which of said first and second sums is smaller.

Claim 22

The invention defined in claim 16 wherein said displacement estimating means is arranged to recursively update each estimate D i using the prior displacement estimate D i-1 , such that D.sub.i -D.sub.i-1 =ε∇I(x-D.sub.i-1,t-τ)·DFD(x,D.sub.i-1) where ∇I(x-D i-1 ,t-τ) is the intensity gradient at location x-D i-1 in the previous frame, DFD(x,D i-1 ) is the intensity difference between location x in the present frame and location x-D i-1 in the previous frame, and ε is a scaling constant.

Claim 23

The invention defined in claim 16 wherein said apparatus further includes: means for variable length encoding each error value only if it exceeds a threshold value, and means for run length encoding the locations of pels for which said error value is encoded.