Encoding adjustments for animation content

• US 8,442,337 B2
• Filed: 04/18/2007
• Issued: 05/14/2013
• Est. Priority Date: 04/18/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Zuo, Xiping, Lin, Chih-Lung
• Primary Examiner(s): Najjar, Saleh
• Assistant Examiner(s): Gao, Shu Chun
• Application Number:
  US11/787,945
• Publication Number:
  US 20080260278A1
• Time in Patent Office: 2,218 Days
• Field of Search: 382/251, 37524001-24029
• US Class Current: 382/251
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/14 Coding unit complexity, e.g. amount of activity or edge presence estimation H04N19/146 takes precedence
• CPC Class Codes: H04N 19/176 the region being a block, e.g. a macroblock
• CPC Class Codes: H04N 19/18 the unit being a set of transform coefficients
• CPC Class Codes: H04N 19/48 using compressed domain processing techniques other than decoding, e.g. modification of transform coefficients, variable length coding [VLC] data or run-length data motion estimation in a transform domain H04N19/547; processing of decoded motion vectors H04N19/513
• CPC Class Codes: H04N 19/61 in combination with predictive coding

Abstract

Techniques and tools are described for encoding animation video. In some embodiments, a video encoder designates animation video for encoding as animation content, which typically involves changing one or more encoder settings or rules to improve encoding performance for the animation content. When the encoder encodes the animation video, the encoder detects edges in the animation video using texture and changes settings for areas that include detected edges so as to improve encoding quality for the areas. In some embodiments, a video encoder adjusts differential quantization rules and quantizes animation video according to the adjusted differential quantization rules.

Claims

1. In a motion-compensated, block-transform video encoder, a computer-implemented method comprising:

designating an animation video for encoding as an animation content, rather than a natural video;

encoding the animation video differently than the natural video, wherein the encoding comprises;

detecting edges in the animation video using texture;

quantizing the animation video, including changing quantization settings of the video encoder to settings that are different than quantization settings for natural video for areas of the animation video that include the detected edges so as to improve encoding quality of the areas;

determining whether the animation content is an intra-coded animation content;

applying a different quantization approach for the intra-coded animation content than for animation content that is not intra-coded;

for the intra-coded animation content, identifying gradient areas with non-zero AC coefficients and adjusting quantization to preserve significant non-zero AC coefficients in some or all of the identified gradient areas, wherein the adjusting quantization includes reducing quantization step size for the gradient areas;

for animation content that is not intra-coded, performing aggressive differential quantization; and

outputting the encoded animation video in an encoded video bit stream.

2. The method of claim 1 wherein the detecting the edges includes creating a map of the edges in a picture of the animation video.

3. The method of claim 1 wherein the detecting the edges comprises:

classifying blocks depending on the texture in the blocks;

dilating preliminary edges resulting from the classifying so as to close gaps in the preliminary edges; and

eroding the dilated edges.

4. The method of claim 1 wherein the detecting the edges includes:

identifying plural blocks as edge blocks; and

identifying a current macroblock as an edge macroblock depending at least in part on the edge blocks and edge status of neighboring macroblocks.

5. The method of claim 4 wherein the edge status of the neighboring macroblocks is indicated at least in part by a count of edge angles among the neighboring macroblocks.

6. The method of claim 1 wherein the changing quantization settings includes narrowing dead zone size of a quantizer for the areas that include the detected edges.

7. The method of claim 1 wherein the changing quantization settings includes decreasing quantization step size for the areas that include the detected edges, compared to other areas of the animation video.

8. The method of claim 1 wherein the changing quantization settings includes reducing or disabling median filtering for the areas that include the detected edges.

9. The method of claim 1 wherein the areas that include detected edges are blocks or macroblocks in an I picture, P picture or B picture of the animation video.

10. A computer-implemented method comprising:

designating an animation video for encoding as an animation content through user input indicating that the animation video includes the animation content rather than natural video, including adjusting differential quantization rules to use for the animation video, the designation as animation video switching an encoder from first quantization settings for the natural video to second quantization settings for the animation video;

encoding the animation video, wherein the encoding comprises quantizing the animation video according to the adjusted differential quantization rules, wherein the encoding includes;

for an intra-coded animation content, identifying gradient areas with non-zero AC coefficients and adjusting quantization to preserve non-zero AC coefficients in some or all of the identified gradient areas, wherein the adjusting quantization includes reducing quantization step size for the gradient areas; and

for animation content that is not intra-coded, performing differential quantization; and

outputting the encoded animation video in an encoded video bit stream.

11. The method of claim 10 wherein the adjusting includes changing bit allocation rules for I pictures and other pictures of the animation video.

12. The method of claim 10 wherein the encoding comprises:

Identifying the gradient areas; and

adjusting quantization for the identified gradient areas to be less coarse.

13. The method of claim 10 wherein the encoding comprises identifying intra-coded areas that include at least one non-zero AC coefficient and decreasing quantization step size for the identified intra-coded areas to preserve one or more of the at least one non-zero AC coefficient.

14. The method of claim 13 wherein the identifying comprises finding blocks or sub-blocks of the gradient areas in I pictures.

15. The method of claim 10 wherein the encoding comprises:

identifying smooth areas and other areas; and

performing aggressive differential quantization for the identified smooth areas relative to the other areas.

16. The method of claim 15 wherein the identifying comprises finding blocks of the smooth areas in P or B pictures, and wherein the aggressive differential quantization decreases quantization step size for macroblocks including the blocks of the smooth areas.

17. The method of claim 10 wherein the adjusting comprises modifying default differential quantization rules or switching from the default differential quantization rules to differential quantization rules adapted for animation content.

18. A system comprising:

an interface for exposing plural encoder settings for user selection, wherein the plural encoder settings include an animation mode setting, which is a different setting than a natural video mode setting; and

a general-purpose video encoder including a motion compensation module implemented in a processor, a frequency transform module, a quantization module and an entropy encoding module, wherein the quantization module incorporates quantization rules adapted for animation content for use when the animation mode is selected, the quantization rules for applying a different quantization approach for an intra-coded and not intra-coded animation content, wherein for the intra-coded animation content the quantization rules include identifying gradient areas with non-zero AC coefficients and adjusting quantization by reducing quantization step size for the gradient areas to preserve significant non-zero AC coefficients in some or all of the identified gradient areas.

19. The system of claim 18 wherein the quantization rules adapted for animation content use different picture-by-picture bit allocations, aggressive texture-based differential quantization rules, edge detection and preservation.

20. The system of claim 18 wherein the quantization rules adapted for animation content include different rules for setting quantization step size, quantization dead zone size and/or median filtering, compared to default quantization rules.