Regions of interest for quality adjustments

• US 8,576,908 B2
• Filed: 07/02/2012
• Issued: 11/05/2013
• Est. Priority Date: 03/30/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Lin, Chih-Lung, Lee, Ming-Chieh, Spears, Stacey, Garrett, Philip Lawrence, Murching, Anil Madhav, Beauford, Amy Jennifer, Holcomb, Thomas W., Zuo, Xiping
• Primary Examiner(s): Dinh, Minh
• Application Number:
  US13/539,730
• Publication Number:
  US 20120269266A1
• Time in Patent Office: 491 Days
• Field of Search: None
• US Class Current: 375/240.05
• CPC Class Codes: H04N 19/124 Quantisation
• CPC Class Codes: H04N 19/147 according to rate distortion criteria rate-distortion as a criterion for motion estimation H04N19/567
• CPC Class Codes: H04N 19/162 User input
• CPC Class Codes: H04N 19/17 the unit being an image region, e.g. an object
• CPC Class Codes: H04N 19/61 in combination with predictive coding

Abstract

Quality settings established by an encoder are adjusted based on information associated with regions of interest (“ROIs”). For example, quantization step sizes can be reduced (to improve quality) or increased (to reduce bit rate). ROIs can be identified and quality settings can be adjusted based on input received from a user interface. An overlap setting can be determined for a portion of a picture that corresponds to an ROI overlap area. For example, an overlap setting is chosen from step sizes corresponding to a first overlapping ROI and a second overlapping ROI, or from relative reductions in step size corresponding to the first ROI and the second ROI. ROIs can be parameterized by information (e.g., using data structures) that indicates spatial dimensions of the ROIs and quality adjustment information (e.g., dead zone information, step size information, and quantization mode information).

Claims

1. A method comprising:

identifying plural regions of interest for a picture, wherein the plural regions of interest comprise a first region of interest that overlaps at least in part with a second region of interest, wherein the first region of interest has first region-of-interest quantization step size information, and wherein the second region of interest has second region-of-interest quantization step size information;

determining an overlap quantization setting for a portion of the picture that corresponds to the overlap, wherein the determining comprises choosing between the first region-of-interest quantization step size information and the second region-of-interest quantization step size information;

adjusting quantization in the portion of the picture that corresponds to the overlap, wherein the adjusting is based at least in part on the overlap quantization setting; and

encoding the picture using the adjusted quantization.

2. The method of claim 1, wherein the first region-of-interest quantization step size information comprises a first absolute quantization step size, wherein the second region-of-interest quantization step size information comprises a second absolute quantization step size, and wherein the choosing comprises choosing the smaller of the first absolute quantization step size and the second absolute quantization step size.

3. The method of claim 1, wherein the first region-of-interest quantization step size information comprises a first relative quantization step size adjustment, and wherein the second region-of-interest quantization step size information comprises a second relative quantization step size adjustment.

4. The method of claim 1, wherein at least some of the plural regions of interest are macroblock-aligned or block-aligned.

5. The method of claim 1, wherein encoding parameters are determined for each region of interest and used to encode the picture.

6. The method of claim 1, wherein the plural regions of interest are rectangular-shaped sub-portions of the picture.

7. The method of claim 1, wherein the picture is in a video stream.

8. A method comprising:

identifying overlapping regions of interest in a picture;

identifying non-overlapping regions of interest in the picture;

calculating a quantization step size for the overlapping regions differently than for non-overlapping regions of interest;

adjusting an encoder setting using at least the calculated quantization step size for the overlapping region; and

encoding the picture using the encoder after the adjusting.

9. The method of claim 8, wherein calculating the quantization step size for overlapping regions comprises:

determining a first quantization step size for a first overlapping region of interest;

determining a second quantization step size for a second overlapping region of interest; and

comparing the first and second quantization step sizes and choosing a lower of the two as the quantization step size of the overlapping region to use to adjust the encoder.

10. The method of claim 8, further comprising:

determining a first dead zone value for a first overlapping region of interest;

determining a second dead zone value for a second overlapping region of interest;

comparing the first dead zone value and the second dead zone value and choosing a lower of the two as the dead zone value for the overlapping region of interest; and

using the chosen dead zone value to adjust an encoder setting.

11. The method of claim 10, further including calculating the first dead zone value using inter block dead zone information and/or intra block dead zone information.

12. The method of claim 8, wherein at least some of the plural regions of interest are macroblock-aligned or block-aligned.

13. The method of claim 8, wherein encoding parameters are determined for each region of interest and used to encode the picture.

14. The method of claim 8, wherein the plural regions of interest are rectangular-shaped sub-portions of the picture.

15. The method of claim 8, wherein the picture is in a video stream.

16. A computer-readable memory having instructions thereon for performing a method, the method comprising:

identifying plural regions of interest for a picture, wherein the plural regions of interest comprise a first region of interest that overlaps at least in part with a second region of interest, wherein the first region of interest has first region-of-interest quantization step size information, and wherein the second region of interest has second region-of-interest quantization step size information;

determining an overlap quantization setting for a portion of the picture that corresponds to the overlap; and

adjusting quantization in the portion of the picture that corresponds to the overlap, wherein the adjusting is based at least in part on the overlap quantization setting.

17. The computer-readable memory of claim 16, wherein the determining comprises choosing between the first region-of-interest quantization step size information and the second region-of-interest quantization step size information.

18. The computer-readable memory of claim 17, wherein the first region-of-interest quantization step size information comprises a first absolute quantization step size, wherein the second region-of-interest quantization step size information comprises a second absolute quantization step size, and wherein the choosing comprises choosing the smaller of the first absolute quantization step size and the second absolute quantization step size.

19. The computer-readable memory of claim 16, wherein the first region-of-interest quantization step size information comprises a first relative quantization step size adjustment, and wherein the second region-of-interest quantization step size information comprises a second relative quantization step size adjustment.

20. The computer-readable memory of claim 16, wherein the plural regions of interest are rectangular-shaped sub-portions of the picture.