Spatially scalable video coding

• US 9,332,274 B2
• Filed: 07/07/2006
• Issued: 05/03/2016
• Est. Priority Date: 07/07/2006
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Inventors: Xu, Ji-Zheng, Xiong, Ruiqin, Wu, Feng
• Primary Examiner(s): VO, TUNG T
• Application Number:
  US11/483,150
• Publication Number:
  US 20080008252A1
• Time in Patent Office: 3,588 Days
• Field of Search: 37524001-24029
• US Class Current: 1/1
• CPC Class Codes: H04N 19/33 in the spatial domain
• CPC Class Codes: H04N 19/59 involving spatial sub-sampling or interpolation, e.g. alteration of picture size or resolution
• CPC Class Codes: H04N 19/61 in combination with predictive coding

Abstract

Video data for a high resolution image unit is coded with regard to both a low resolution reference image unit and a high resolution reference image unit. In an example encoding implementation, both low pass information and high pass information of residue data for a current image are generated. In an example decoding implementation, a current image is reconstructed by synthesizing both low pass information and high pass information for the reconstructed image.

Claims

1. A method for encoding a current image, the method comprising:

generating low pass information of residue data for the current image based on low pass information for the current image;

ascertaining high pass information for the current image responsive to the current image based at least in part on calculating a difference between the current image and up-sampled information for the current image;

ascertaining high pass information of prediction data for the current image by applying prediction data for the current image to a high-pass information determining processor that mixes the prediction data with a down-sampled then up-sampled version of the prediction data to generate the high pass information of prediction data for the current image; and

generating high pass information of residue data for the current image based at least in part on calculating a difference between the high pass information for the current image and the high pass information of prediction data for the current image.

2. The method as recited in claim 1, wherein the generating low pass information of residue data for the current image based on low pass information for the current image comprises:

calculating a difference between the low pass information for the current image and low pass information of prediction data for the current image.

3. The method as recited in claim 2, further comprising:

creating the low pass information for the current image by down-sampling the current image; and

determining the low pass information of prediction data for the current image by applying a motion compensation process to low pass information for a reference image.

4. The method as recited in claim 1, wherein the ascertaining high pass information for the current image responsive to the current image comprises:

creating the low pass information for the current image by down-sampling the current image; and

creating the up-sampled information for the current image by up-sampling the low pass information for the current image.

5. The method as recited in claim 1, wherein the ascertaining high pass information for the current image responsive to the current image comprises:

ascertaining the high pass information for the current image responsive to a reconstructed version of the low pass information for the current image.

6. The method as recited in claim 1, wherein the ascertaining high pass information of prediction data for the current image responsive to prediction data for the current image comprises:

creating low pass information of prediction data for the current image by down-sampling the prediction data for the current image to create the down-sample version of the prediction data;

creating up-sampled information of prediction data for the current image by up-sampling the low pass information of prediction data for the current image to create the up-sampled version of the prediction data; and

calculating, by the high-pass determining processor, a difference between the prediction data for the current image and the up-sampled information of prediction data for the current image to create the high pass information of prediction data.

7. The method as recited in claim 6, further comprising:

determining the prediction data for the current image by applying a motion compensation process to a reference image.

8. The method as recited in claim 1, wherein the ascertaining high pass information of prediction data for the current image responsive to prediction data for the current image comprises:

convolving a filter and the prediction data for the current image, wherein taps of the filter are acquired by minimizing an energy of residue signals given a constraint on a number of the taps of the filter.

9. The method as recited in claim 1, further comprising:

transmitting the low pass information of residue data for the current image; and

transmitting the high pass information of residue data for the current image that corresponds to an identified resolution level of a destination device.

10. A method for decoding a current image, the method comprising:

generating low pass information for a reconstructed image based on low pass information of residue data for the current image;

determining high pass information of prediction data by applying prediction data for the current image to a high-pass information determining processor that mixes the prediction data with a down-sampled then up-sampled version of the prediction data to generate the high pass information of prediction data;

generating high pass information for the reconstructed image based on high pass information of residue data for the current image and based on the high pass information of prediction data; and

producing a reconstructed current image by synthesizing the low pass information for the reconstructed image and the high pass information for the reconstructed image.

11. The method as recited in claim 10, wherein the generating low pass information for a reconstructed image based on low pass information of residue data for the current image comprises:

combining the low pass information of residue data for the current image and low pass information of prediction data, the low pass information of prediction data determined from low pass information for a reference image.

12. The method as recited in claim 10, wherein the generating high pass information for the reconstructed image based on high pass information of residue data for the current image comprises:

combining the high pass information of residue data for the current image and the high pass information of prediction data.

13. The method as recited in claim 12, further comprising:

determining the prediction data for the current image from a reference image at a high resolution and a motion field at the high resolution.

14. The method as recited in claim 10, wherein the producing a reconstructed current image by synthesizing the low pass information for the reconstructed image and the high pass information for the reconstructed image comprises:

up-sampling the low pass information for the reconstructed image to create up-sampled low pass information for the reconstructed image; and

combining the up-sampled low pass information for the reconstructed image and the high pass information for the reconstructed image to produce the reconstructed current image.

15. A device comprising:

a processor;

a memory coupled to the processor, wherein instructions stored in the memory program the processor to;

code a current image unit at a given resolution level using first prediction data determined at the given resolution level and second prediction data determined at a lower resolution level, the coding comprising;

applying the first prediction data to a high pass information determining processor that mixes the first prediction data with a down-sampled then up-sampled version of the first prediction data to generate high pass information of prediction data for the current image;

determining high pass information for the current image based at least in part on calculating a difference between the current image and up-sampled information for the current image;

generating high pass information of residue data for the current image unit based at least in part on the high pass information for the current image and the high pass information of prediction data.

16. The device as recited in claim 15, wherein the second prediction data determined at the lower resolution level is determined from low pass information for a reference image having a time that corresponds to a time of the current image unit.

17. The device as recited in claim 15, wherein the first prediction data determined at the given resolution level is determined from a reference image having a time that differs from a time of the current image unit.

18. The device as recited in claim 15, wherein the video coder generates low pass information of residue data based on the second prediction data determined at the lower resolution level and high pass information of residue data based on the first prediction data determined at the given resolution level.

19. The device as recited in claim 15, wherein the video coder synthesizes a reconstructed version of the current image from low pass information for a reconstructed image that is based on the second prediction data determined at the lower resolution level and high pass information for the reconstructed image that is based on the first prediction data determined at the given resolution level.