Motion estimation and encoding of video signals in the transform domain

• US 4,245,248 A
• Filed: 04/04/1979
• Issued: 01/13/1981
• Est. Priority Date: 04/04/1979
• Status: Expired due to Term

• Current Assignee: Bell Telephone Laboratories, Inc. (Nokia Corporation)
• Original Assignee: Bell Telephone Laboratories, Inc. (Nokia Corporation)
• Inventors: Netravali, Arun N., Stuller, John A.
• Primary Examiner(s): Griffin, Robert L.
• Assistant Examiner(s): Orsino, Jr., Joseph A.
• Application Number:
  US06/027,053
• Time in Patent Office: 650 Days
• Field of Search: 358/133, 358/135, 358/136, 358/105, 364/515, 364/521
• US Class Current: 348/400.1
• CPC Class Codes: H04N 19/619 the transform being operated outside the prediction loop

Abstract

Video signals are encoded (FIG. 5) using motion compensated prediction which operates on a transform domain representation of the signal. The displacement estimate used for compensation is recursively updated, so that a feedback path exists between the output (205) of the displacement calculator and one of its output (209). The update term is also computed in the transform domain. A decoder (FIG. 6) uses the same prediction technique to recover the original picture.

Claims

1. A method of encoding a video signal including a series of samples representing the intensities of elements of a picture, including the steps of:

(a) linearly transforming the intensity values for a first block of picture elements (pels) to a corresponding set of coefficient values,

(b) predicting the values of said coefficients, and

(c) encoding an error signal indicative of the difference between ones of said coefficient values and the predicted values thereof,

wherein said predicting step is CHARACTERIZED BY

(d) estimating the displacement of objects in said picture between a previous frame and the present frame, and

(e) predicting said values using the coefficients of a second block of pels in said previous frame which is displaced from said first block by said estimate.

2. The invention defined in claim 1 wherein said estimating step includes recursively updating each estimate as a function of the previous estimate.

3. The method defined in claim 2 wherein said updating step includes forming an update term proportional to the product of (a) said difference between a coefficient value and the corresponding predicted value thereof, and (b) a coefficient gradient of said second block of pels.

4. The method defined in claim 3 wherein said encoding step includes:

(1) comparing said difference to a threshold value, and

(2) encoding said difference only if said threshold value is exceeded.

5. A method of encoding a series of intensity values each representing an element of a picture, comprising the steps of:

(a) transforming the intensity values for a block of pels to a series of coefficients representative thereof,

(b) predicting the value of said coefficients based upon the value of coefficients for a block of pels in a previous frame of said picture, and

(c) encoding a prediction error signal representing the difference between the actual and predicted values of ones of said coefficients,

wherein said predicting step is CHARACTERIZED BY

(d) selecting said block of pels in said previous frame in accordance with an estimate of the displacement of objects in said picture occurring between said previous frame and the present frame.

6. The method defined in claim 5 wherein said selecting step includes updating each displacement estimate as a function of the previous displacement estimate.

7. The method defined in claim 6 wherein said updating step includes:

(1) storing each displacement estimate, and

(2) combining each stored estimate with a term proportional to the product of said difference between actual and predicted values and a coefficient gradient of said block of pels in said previous frame.

8. The method defined in claim 7 wherein said coefficient gradient is obtained by the steps of:

(1) forming line and element differences of the pels in said previous frame block, and

(2) transforming said differences to a series of coefficients representative thereof according to the transformation of step (a).

9. The method defined in claim 8 wherein said transforming step comprises a Hadamard transformation.

10. A method of encoding a video signal which includes a series of samples each indicating the intensity of an element of a picture, including the steps of:

(1) converting the intensity values for a first block of pels to a series of coefficient values through a linear transformation,

(2) predicting the values of each of said coefficients based upon stored intensity values for elements in a previous frame of said picture, and

(3) encoding said signal as a function of the difference between predicted and actual coefficient values,

CHARACTERIZED BY

(4) estimating the displacement of objects in said picture occurring in the period between said prior frame and the present frame, and

(5) selecting stored values for a second block of pels for said prediction based upon said displacement estimate.

11. The method defined in claim 10 wherein said estimating step includes updating each estimate each time said signal is encoded, as a function of the prior estimate.

12. A method of encoding a video signal which includes a series of samples each indicating the intensity of an element of a picture, including the steps of:

(1) converting the intensity values for a first block of pels to a series of coefficient values through a linear transformation.

(2) predicting the values of each of said coefficients based upon stored intensity values for elements in a previous frame of said picture, and

(3) encoding said signal as a function of the difference between predicted and actual coefficient values,

CHARACTERIZED BY

(4) estimating the displacement of objects in said picture occurring in the period between said prior frame and the present frame, said estimating step including updating each estimate each time said signal is encoded, as a function of the prior estimate, and

(5) selecting stored values for said prediction based upon said displacement estimate, said selecting step including interpolating coefficient values derived from said previous frame to yield the coefficient values for a second block of pels displaced from said first block by said displacement estimate.

13. The method of claim 12 wherein said update is proportional to the product of (a) said difference between predicted and actual coefficient values and (b) a coefficient value calculated using element and line differences for pels in said second block.

14. The method defined in claim 13 wherein said encoding step includes comparing said difference to a predetermined threshold value.

15. Apparatus for encoding a series of intensity values each representing an element of a picture, comprising:

(a) means (101) transforming the intensity values for a block of pels to a series of coefficients representative thereof,

(b) means (200) for predicting the value of said coefficients based upon the value of coefficients for a block of pels in a previous frame of said picture, and

(c) means (103) for encoding a prediction error signal representing the difference between the actual and predicted values of ones of said coefficients,

wherein said predicting means is CHARACTERIZED BY

(d) means (203,

     210) for selecting said block of pels in said previous frame in accordance with an estimate of the displacement of objects in said picture occurring between said previous frame and the present frame.

16. The invention defined in claim 15 wherein said selecting means includes means (208) for updating each displacement estimate as a function of the previous displacement estimate.

17. The invention defined in claim 16 wherein said selecting means includes:

(1) means (206) for storing each displacement estimate, and

(2) means (207) for combining each stored estimate with a term proportional to the product of said difference between actual and predicted values and a coefficient gradient in said block of pels in said previous frame.

18. Apparatus for encoding a video signal which includes a series of samples each indicating the intensity of an element of a picture, including:

(1) means (101) for converting the intensity values for a first block of pels to a series of coefficient values through a linear transformation,

(2) means (200) for predicting the values of each of said coefficients based upon stored intensity values for elements in a previous frame of said picture, and

(3) means (103) for encoding said signal as a function of the difference between predicted and actual coefficient values,

CHARACTERIZED BY

(4) means (210) for estimating the displacement of objects in said picture occurring in the period between said prior frame and the present frame, and

(5) means (203) for selecting stored values for a second block of pels for said prediction based upon said displacement estimate.

19. The invention defined in claim 18 wherein said estimating means includes means (208) for updating each estimate each time said signal is encoded, as a function of the prior estimate.

20. Apparatus for encoding a video signal which includes a series of samples each indicating the intensity of an element of a picture, including:

(1) means (101) for converting the intensity values for a first block of pels to a series of coefficient values through a linear transformation,

(2) means (200) for predicting the values of each of said coefficients based upon stored intensity values for elements in a previous frame of said picture, and

(3) means (103) for encoding said signal as a function of the difference between predicted and actual coefficient values,

CHARACTERIZED BY

(4) means (210) for estimating the displacement of objects in said picture occurring in the period between said prior frame and the present frame, wherein said estimating means includes means (208) for updating each estimate, each time said signal is encoded, as a function of the prior estimate, and

(5) means (203) for selecting stored values for said prediction based upon said displacement estimate, wherein said selecting means includes means (536) for interpolating coefficient values derived from said previous frame to yield coefficient values for a second block of pels displaced from said first block by said displacement estimate.

21. The invention of claim 20 wherein said updating means (208) is arranged to form an output proportional to the product of (a) said difference between predicted and actual coefficient values and (b) a coefficient value calculated using element and line differences for pels in said second block.

22. The invention defined in claim 21 wherein said encoding means includes means (510) for comparing said difference to a predetermined threshold value.