METHOD FOR MOTION ESTIMATION

• US 20100165210A1
• Filed: 03/28/2007
• Published: 07/01/2010
• Est. Priority Date: 03/29/2006
• Status: Active Grant

• Current Assignee: Sony Deutschland GmbH (Sony Group Corp.)
• Original Assignee: Sony Deutschland GmbH (Sony Group Corp.)
• Inventors: Tappermann, Joerg, Incesu, Yalcin
• Granted Patent: US 8,724,703 B2
• Time in Patent Office: Days
• US Class Current: 348/699
• CPC Class Codes: H04N 19/533 Motion estimation using multistep search, e.g. 2D-log search or one-at-a-time search [OTS]

Abstract

A method for motion estimation with respect to a sequence of images. Based on predictor vectors respective updates are generated. The updates are distributed over a plurality of tables. Only a single table of updates is applied to a given predictor vector to generate a limited set of candidate vectors only.

Claims

1-20. -20. (canceled)

21. A method for motion estimation with respect to a sequence of images, comprising:

generating respective updates based on predictor vectors;

distributing the updates over a plurality of tables; and

applying only a single table of updates to a given predictor vector to generate a limited set of candidate vectors only.

22. A method according to claim 21, comprising:

(a) providing digital image data representing a sequence of images;

(b) providing one or a plurality of predictor vectors;

(c) generating updates for the predictor vectors;

(d) generating for the updates a candidate vector, thereby providing a plurality of candidate vectors by applying selected updates to a given predictor vector;

(e) selecting from the plurality of candidate vectors a distinct candidate vector that fulfils a given matching criterion as a motion vector candidate;

(f) wherein the generating (c), generating (d), and generating (e) are performed recursively until a given termination criterion is fulfilled;

(g) wherein, as a result of each recursion, the predictor vectors or a subset thereof are updated by providing respective updates;

(h) wherein the updates are distributedly organized and stored within a plurality of update tables;

(i) wherein as a consequence each update table contains less than a complete number of updates; and

(j) wherein in the generating (d) generating the candidate vectors only the updates of a single update table are applied to a given predictor vector.

38. A system for motion estimation, comprising:

means to perform the method for motion estimation according to claim 21.

39. A computer program product, comprising:

computer program means to perform a method for motion estimation according to claim 21 when executed on a computer or a digital signal processor.

40. A computer readable storage medium, comprising:

a computer program product according to claim 39.

23. A method for motion estimation with respect to a sequence of images, comprising:

(a) providing digital image data representing a sequence of images;

(b) providing one or a plurality of predictor vectors;

(c) generating updates for the predictor vectors;

(d) generating for the updates a candidate vector, thereby providing a plurality of candidate vectors by applying selected updates to a given predictor vector;

(e) selecting from the plurality of candidate vectors a distinct candidate vector that fulfils a given matching criterion as a motion vector candidate;

(f) wherein the generating (c), generating (d), and generating (e) are performed recursively until a given termination criterion is fulfilled;

(g) wherein, as a result of each recursion, the predictor vectors or a subset thereof are updated by providing respective updates;

(h) wherein the updates are distributedly organized and stored within a plurality of update tables;

(i) wherein as a consequence each update table contains less than a complete number of updates; and

(j) wherein in the generating (d) generating said candidate vectors only the updates of a single update table are applied to a given predictor vector.

24. A method according to claim 23,

wherein the plurality of update tables is organized in a plurality of table sets;

and

wherein the number of table sets is less than the number of tables.

25. A method according to claim 24,

wherein each table set contains an equal number or almost equal number of update tables.

26. A method according to claim 23,

wherein the updates are equally or almost equally distributed over the entirety of update tables.

27. A method according to claim 23,

wherein each update contains or represents an X component update value and an Y component update value.

28. A method according to claim 23,

wherein the update tables and the table set are stored in a memory.

29. A method according to claim 28,

wherein the memory is dynamically configurable and/or dynamically configured during a run-time.

30. A method according to claim 29,

wherein the update tables are dynamically reconfigured with respect to their content;

and

wherein thereby given image sequence properties are fulfilled.

31. A method according to claim 30,

wherein the image sequence properties are one or a plurality of small motion, strong motion, and rotation.

32. A method according to claim 23,

wherein the (b) providing one or a plurality of predictor vectors is included in the process (f) of recursion.

33. A method according to claim 23,

wherein for each predictor vector about 50 updates are generated, which are distributed over about 5 update tables.

34. A method according to claim 23,

wherein possible positive, negative, short, long, integer, and/or fractional update values are homogenously distributed over the update tables to reduce a spatial correlation.

35. A method according to claim 23,

wherein estimation is started for a new vector row with a table index that is different from a table index of the previously estimated vector row.

36. A method according to claim 23,

wherein estimation of a new vector field is started with a table index that is different from a table index of a previously estimated vector field.

37. A method according to claim 23,

wherein selection of an update and/or of a table set is performed by using a motion model classifier.