Adaptive error detection for MPEG-2 error concealment

Patent metadata

Field	Value
Publication number	US-9848209-B2
Application number	US-6156708-A
Country	US
Kind code	B2
Filing date	Apr 2, 2008
Priority date	Apr 2, 2008
Publication date	Dec 19, 2017
Grant date	Dec 19, 2017

Abstract

A decoder which can detect errors in MPEG-2 coefficient blocks can identify syntactically-correct blocks which have out-of-bounds coefficients. The decoder computes coefficient bounds based on quantization scalers and quantization matrices and compares these to coefficient blocks during decoding; if a block has out-of-bounds coefficients, concealment is performed on the block. In a decoder implemented all in software, coefficient bounds checking is performed on iDCT coefficients against upper and lower bounds in a spatial domain. In a decoder which performs iDCT in hardware, DCT coefficients are compared to an upper energy bound.

First claim

We claim:

1. A method for determining in a video decoder if a video block contains an error, the method comprising:
receiving compressed video information in a bitstream for the video block;
computing a coefficient bound associated with coefficients of the video block using per transform coefficient weights based at least in part on a quantization matrix for the video block, the per transform coefficient weights depending on the quantization matrix and indicating coarseness of quantization during compression of the video information;
detecting errors in the compressed video information in the bitstream for the video block by:
checking whether any of the coefficients of the video block are out of bounds of a range for correct decoding of the video block, including comparing at least some of the coefficients of the video block to the coefficient bound; and
determining that one or more of the coefficients of the video block are in error based on the comparison to the coefficient bound.

2. The method of claim 1 , wherein the coefficient bound is an upper bound.

3. The method of claim 1 , wherein the coefficient bound is a lower bound.

4. The method of claim 1 , wherein the coefficients of the video block are dequantized coefficients, and wherein:
the video block comprises quantized coefficients in the bitstream for the video block;
the method further comprises dequantizing the quantized coefficients of the video block using the quantization matrix to derive the dequantized coefficients in a frequency domain;
the coefficient bound is an energy bound in the frequency domain; and
the dequantized coefficients are compared to the energy bound in the frequency domain.

5. The method of claim 4 , wherein the coefficient bound is an energy bound for the video block.

6. The method of claim 4 , wherein the coefficients are compared in the frequency domain before being input into a hardware-assisted inverse transform process.

7. The method of claim 6 , wherein the hardware-assisted inverse transform process uses a video acceleration interface.

8. The method of claim 1 , wherein the coefficients of the video block are spatial coefficients, and wherein:
the video block comprises quantized coefficients in the bitstream for the video block;
the method further comprises:
dequantizing the quantized coefficients of the video block using the quantization matrix to derive dequantized coefficients in a frequency domain; and
inverse transforming the dequantized coefficients to derive the spatial coefficients of the video block;

the coefficient bound is a spatial coefficient bound; and
the spatial coefficients are compared to the coefficient bound in a spatial domain.

9. The method of claim 1 , further comprising, after determining that one or more of the coefficients are in error, concealing the video block.

10. The method of claim 1 , wherein the video block is syntactically correct.

11. The method of claim 1 , wherein the video block is an intra block.

12. The method of claim 1 , wherein the video block is an inter block.

13. The method of claim 1 , wherein the coefficient bound is computed before receiving the compressed video information for the video block.

14. The method of claim 1 , wherein the quantization matrix indicates, for each of multiple coefficient frequencies of the video block, a weight for inverse quantization.

15. The method according to claim 1 , wherein the computing of the coefficient bound is further based on quantization error in the spatial domain during compression of the video information.

16. A method for detecting an error in a syntactically-correct MPEG-2 bitstream in a video decoder, the method comprising:
receiving compressed video information for a syntactically-correct block of an MPEG-2 bitstream, the block associated with a quantization matrix and a quantization scaler for scaling quantized transform coefficients of the block;
computing an upper coefficient bound and a lower coefficient bound for a block of spatial coefficients from per transform coefficient weights based on the quantization matrix and the quantization scaler, the per transform coefficient weights depending on the quantization matrix and indicating coarseness of quantization during compression of the video information;
performing dequantization and inverse discrete cosine transform on the block of the MPEG-2 bitstream to generate the block of spatial coefficients;
detecting errors in the compressed video information in the bitstream for the block by checking whether any spatial coefficients in the block are out of bounds of a range for correct decoding of the block, wherein the checking includes comparing the spatial coefficients in the block with the upper coefficient bound and the lower coefficient bound; and
when at least one of the spatial coefficients in the block falls outside of the upper coefficient bound or the lower coefficient bound, concealing the block during display.

17. The method of claim 16 , wherein the quantization matrix indicates, for each of multiple coefficient frequencies of the video block, a weight for inverse quantization.

18. The method according to claim 16 , wherein the computing of the upper coefficient bound and the lower coefficient bound is further based on quantization error in the spatial domain during compression of the video information.

19. One or more computer-readable media selected from the group consisting of: volatile memory, non-volatile memory, magnetic disk storage, magnetic tape storage, CD-ROM, DVD, and Blu-Ray disc, the one or more computer-readable media storing computer-executable instructions which, when executed, cause a computing device programmed thereby to perform operations to determine in a video decoder if a video block in a video bitstream contains an error, the operations comprising:
receiving compressed video information in the video bitstream for the video block;
dequantizing the video block using a quantization matrix;
computing one or more coefficient bounds associated with coefficients of the video block using per transform coefficient weights based at least in part on the quantization matrix;
detecting errors in the compressed video information in the bitstream for the video block by:
checking whether any of the coefficients of the video block are out of bounds of a range for correct decoding of the video block, including comparing at least some of the coefficients of the video block to the one or more coefficient bounds; and
determining that one or more of the coefficients of the video block are in error based on the comparison to the one or more coefficient bounds; and

concealing the block when rendering.

20. The one or more computer-readable media of claim 19 , wherein:
the one or more coefficient bounds comprise an upper energy bound in a frequency domain; and
the coefficients are compared to the upper energy bound in the frequency domain.

21. The one or more computer-readable media of claim 20 , wherein the coefficients are compared in the frequency domain before being input into a hardware-assisted inverse transform process.

22. The one or more computer-readable media of claim 19 , wherein the coefficients of the video block are spatial coefficients, and wherein:
the operations further comprise inverse transforming dequantized coefficients of the video block to derive the spatial coefficients of the video block;
the one or more coefficient bounds are an upper spatial coefficient bound a

Assignees

• Ji Gang
• Wu Yongjun
• Folta Florin
• Thumpudi Naveen
• Microsoft Technology Licensing Llc

Inventors

• Ji Gang
• Wu Yongjun
• Folta Florin
• Thumpudi Naveen

Classifications

• H04N19/134

  characterised by the element, parameter or criterion affecting or controlling the adaptive coding · CPC title

• H04N19/176

  the region being a block, e.g. a macroblock · CPC title

• H04N19/18

  the unit being a set of transform coefficients · CPC title

• H04N19/46

  Embedding additional information in the video signal during the compression process (H04N19/517, H04N19/68, H04N19/70 take precedence) · CPC title

• H04N19/895 Primary

  in combination with error concealment · CPC title