Encoder circuit and encoding method

• US 10,448,047 B2
• Filed: 08/24/2015
• Issued: 10/15/2019
• Est. Priority Date: 09/01/2014
• Status: Active Grant

• Current Assignee: Socionext Inc.
• Original Assignee: Socionext Inc.
• Inventors: Nakaishi, Hidenori
• Primary Examiner(s): Atala, Jamie J
• Assistant Examiner(s): Billah, Masum
• Application Number:
  US14/834,294
• Publication Number:
  US 20160065985A1
• Time in Patent Office: 1,513 Days
• Field of Search: 375240
• CPC Class Codes: H04N 19/533 Motion estimation using multistep search, e.g. 2D-log search or one-at-a-time search [OTS]
• CPC Class Codes: H04N 19/56 Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search
• CPC Class Codes: H04N 19/57 Motion estimation characterised by a search window with variable size or shape

Abstract

An encoder circuit includes a reduction circuit, a statistical processing circuit, a first motion search circuit, and a second motion search circuit. The reduction circuit generates reduced images of multiple patterns with respect to each of macroblocks. Each of frame images included in video is divided into the macroblocks. The statistical processing circuit performs statistical processing on motion vectors detected with respect to each of the macroblocks. The first motion search circuit selects one of the reduced images generated with respect to the frame image of a reference frame, based on the result of the statistical processing, and searches a first search range in the selected one of the reduced images so as to detect a first motion vector. The second motion search circuit searches a second search range set in the reference frame based on the first motion vector so as to detect a second motion vector.

Claims

1. An encoder circuit, comprising:

a reduction circuit configured to generate a plurality of reduced images in accordance with a plurality of pixel reduction patterns, the plurality of reduced images being generated for each of macroblocks, the plurality of pixel reduction patterns having a same reduction ratio in each of a horizontal direction and a vertical direction of the macroblock, each of a plurality of frame images included in video being divided into the macroblocks;

a statistical processing circuit configured to perform statistical processing on motion vectors detected with respect to each of the macroblocks;

a first motion search circuit configured to select one of the plurality of reduced images of the plurality of pixel reduction patterns generated by the reduction circuit with respect to the frame image of a reference frame, based on a result of the statistical processing, and search a first search range in the selected one of the plurality of reduced images so as to detect a first motion vector; and

a second motion search circuit configured to search a second search range set in the reference frame based on the first motion vector so as to detect a second motion vector, wherein

each of the macroblocks is a pixel matrix, and

the reduction circuit is configured to generate the plurality of reduced images for each of the macroblocks by reducing pixels included in the pixel matrix in accordance with the plurality of pixel reduction patterns having the same reduction ratio with a plurality of set directions that differ in a ratio of a component in a first direction of arrangement and a component in a second direction of arrangement perpendicular to the first direction of arrangement, and

the first motion search circuit is configured to select the one of the plurality of reduced images of the plurality of pixel reduction patterns in accordance with a ratio of a first component and a second component of a statistical motion vector, the statistical motion vector being obtained as the result of the statistical processing, the first component being in a first direction of arrangement and the second component being in a second direction of arrangement perpendicular to the first direction of arrangement.

2. The encoder circuit as claimed in claim 1, wherein the reduction circuit is configured to generate each of the plurality of reduced images by setting first lines and second lines alternately in the set direction, the first lines and second lines being perpendicular to the set direction, respectively, and eliminating every other pixel on the first lines and eliminating each of the pixels on the second lines.

3. The encoder circuit as claimed in claim 1, wherein the statistical processing circuit is configured to perform the statistical processing on the motion vectors of neighboring macroblocks of one of the macroblocks that is a target of processing in one of the frame images that is a target of processing.

4. The encoder circuit as claimed in claim 3, wherein the statistical processing circuit is configured to perform the statistical processing on the motion vectors detected in the one of the frame images and the motion vectors detected in a reference frame subsequent to the one of the frame images.

5. The encoder circuit as claimed in claim 1, wherein the statistical processing circuit is configured to perform the statistical processing on the motion vectors of, neighboring macroblocks of one of the macroblocks that is a target of processing in one of the frame images that is a target of processing and the motion vectors of the macroblocks in a reference region in the reference frame, the reference region including a macroblock spatially at a same position as the one of the macroblocks.

6. An encoding method, comprising:

generating a plurality of reduced images in accordance with a plurality of pixel reduction patterns, the plurality of reduced images being generated for each of macroblocks, the plurality of pixel reduction patterns having a same reduction ratio in each of a horizontal direction and a vertical direction of the macroblock, each of a plurality of frame images included in video being divided into the macroblocks;

performing statistical processing on motion vectors detected with respect to each of the macroblocks;

selecting one of the plurality of reduced images of the plurality of pixel reduction patterns generated with respect to the frame image of a reference frame, based on a result of the statistical processing;

searching, by a circuit, a first search range in the selected one of the plurality of reduced images so as to detect a first motion vector; and

searching, by the circuit, a second search range set in the reference frame based on the first motion vector so as to detect a second motion vector, wherein

each of the macroblocks is a pixel matrix, and

said generating the reduced images generates the plurality of reduced images for each of the macroblocks by reducing pixels included in the pixel matrix in accordance with the plurality of pixel reduction patterns having the same reduction ratio with a plurality of set directions that differ in a ratio of a component in a first direction of arrangement and a component in a second direction of arrangement perpendicular to the first direction of arrangement, and

said selecting selects the one of the plurality of reduced images of the plurality of pixel reduction patterns in accordance with a ratio of a first component and a second component of a statistical motion vector the statistical motion vector being obtained the result statistical processing, component being in a first direction of arrangement and the second component being in a second direction of arrangement perpendicular to the first direction of arrangement.

7. The encoding method as claimed in claim 6, wherein said generating the reduced images generates each of the plurality of reduced images by setting first lines and second lines alternately in the set direction, the first lines and second lines being perpendicular to the set direction, respectively, and eliminating every other pixel on the first lines and eliminating each of the pixels on the second lines.

8. The encoding method as claimed in claim 6, wherein said performing the statistical processing performs the statistical processing on the motion vectors of neighboring macroblocks of one of the macroblocks that is a target of processing in one of the frame images that is a target of processing.

9. The, encoding method as claimed in claim 8, wherein said performing the statistical processing performs the statistical processing on the motion vectors detected in the one of the frame images and the motion vectors detected in a reference frame subsequent to the one of the frame images.

10. The encoding method as claimed in claim 6, wherein said performing the statistical processing performs the statistical processing on the motion vectors of neighboring macroblocks of one of the macroblocks that is a target of processing in one of the frame images that is a target of processing and the motion vectors of the macroblocks in a reference region in the reference frame, the reference region including a macroblock spatially at a same position as the one of the macroblocks.