Gradient slope detection for video compression

• US 20070237237A1
• Filed: 04/07/2006
• Published: 10/11/2007
• Est. Priority Date: 04/07/2006
• Status: Abandoned Application

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Lin, Chih-Lung, Chang, Cheng
• Application Number:
  US11/400,739
• Publication Number:
  US 20070237237A1
• Time in Patent Office: Days
• US Class Current: 375/240.210
• 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 19/61 in combination with predictive coding

Abstract

A video encoder detects gradient slope content in a video picture by checking for gradient directions for plural regions (e.g., 16×16 macroblocks) in the video picture comprising plural pixels. The encoder processes the gradient slope content differently than other kinds of content in the picture. The encoder can down-sample the video picture and check the down-sampled video picture for gradient slope content. The encoder can find smooth blocks and then analyze only smooth blocks for gradient slope characteristics. A video encoder detects gradient slope content in a video picture and compresses the gradient slope content by performing differential quantization on the gradient slope content to reduce contouring artifacts in the video picture. For example, the encoder uses a selected quantization step size for the gradient slope content, where the selected quantization step size for the gradient slope content is smaller than a quantization step size for non-gradient slope content.

Claims

1. In a video encoder, a method comprising:

detecting gradient slope content in a video picture;

processing the gradient slope content in the video picture differently than other kinds of content in the video picture; and

outputting results of the processing;

wherein the detecting gradient slope content comprises checking for gradient directions for plural regions in the video picture comprising plural pixels.

2. The method of claim 1 wherein each of the plural regions is a 16×

16 macroblock.

3. The method of claim 1 wherein the detecting gradient slope content further comprises down-sampling the video picture and checking the down-sampled video picture for gradient slope content.

4. The method of claim 1 wherein the detecting gradient slope content further comprises analyzing a texture map.

5. The method of claim 4 wherein the texture map classifies each of plural blocks in the video picture as smooth, textured, or edge.

6. The method of claim 1 wherein the detecting gradient slope content further comprises analyzing only smooth blocks for gradient slope characteristics.

7. The method of claim 1 further comprising generating a gradient slope decision mask.

8. The method of claim 7 wherein the gradient slope decision mask indicates a gradient slope decision for each of the plural regions.

9. One or more computer-readable media having stored thereon computer-executable instructions to cause a computer to perform the method of claim 1.

10. In a video encoder, a method comprising:

detecting gradient slope content in a video picture;

compressing the gradient slope content in the video picture; and

outputting results of the compressing;

wherein the compression of the gradient slope content comprises performing differential quantization on the gradient slope content to reduce contouring artifacts in the video picture.

11. The method of claim 10 wherein the differential quantization comprises:

selecting a quantization step size for the gradient slope content, wherein the compression of the gradient slope content uses the selected quantization step size for the gradient slope content, and wherein the selected quantization step size for the gradient slope content is smaller than a quantization step size for non-gradient slope content in the video picture.

12. The method of claim 10 further comprising:

compressing non-gradient slope content in the video picture;

wherein the compression of the gradient slope content differs from the compression of the non-gradient slope content.

13. The method of claim 10 further comprising generating a gradient consistency mask.

14. The method of claim 13 wherein the generating the gradient consistency mask comprises performing morphological operations on the gradient consistency mask.

15. One or more computer-readable media having stored thereon computer-executable instructions to cause a computer to perform the method of claim 10.

16. In a video encoder, a method comprising:

down-sampling a video picture comprising plural blocks;

for each of plural regions of the down-sampled video picture;

obtaining texture classification information for the region;

detecting plural pixel gradients for the region; and

calculating a gradient for the region based at least in part on the plural pixel gradients;

for each of the plural regions of the down-sampled video picture, performing a gradient consistency check for the region;

making gradient slope decisions for the video picture based at least in part on the consistency checks; and

compressing the video picture based at least in part on the gradient slope decisions.

17. The method of claim 16 wherein the texture classification information for the region is obtained from a texture map for the plural blocks of the video picture.

18. The method of claim 16 wherein the making the gradient slope decisions comprises analyzing a gradient consistency mask.

19. The method of claim 18 wherein the analyzing the gradient consistency mask comprises bucket voting.

20. One or more computer-readable media having stored thereon computer-executable instructions to cause a computer to perform the method of claim 16.