Pixel-based intra prediction for coding in HEVC

Patent metadata

Field	Value
Publication number	US-9516349-B2
Application number	US-201213546848-A
Country	US
Kind code	B2
Filing date	Jul 11, 2012
Priority date	Jul 12, 2011
Publication date	Dec 6, 2016
Grant date	Dec 6, 2016

Abstract

An apparatus comprising a processor configured to receive a video frame comprising a current block, wherein the current block comprises a plurality of original pixels, compute a plurality of reconstructed pixels based on the original pixels, and use an intra prediction mode to generate a set of prediction pixels based on a set of reconstructed pixels in the plurality of reconstructed pixels.

First claim

What is claimed is:

1. An apparatus comprising:
a memory;
a processor coupled to the memory and configured to:
receive a video frame comprising a current block, wherein the current block comprises a plurality of original pixels;
compute a plurality of reconstructed pixels based on the original pixels; and
use an intra prediction mode to generate sets of prediction pixels in sequence one of vertically and horizontally, but not both, for the current block, wherein the sets of prediction pixels are each smaller than the current block and based on a set of reconstructed pixels in the plurality of reconstructed pixels, wherein a set of internal reference pixels from within the current block immediately precedes each set of prediction pixels generated in sequence after an initial set of prediction pixels such that a prediction distance remains one pixel regardless of a size of the current block,
select the intra prediction mode from a plurality of available intra prediction modes;
compute a difference between each of the set of prediction pixels and each of a corresponding set of original pixels in the plurality of original pixels to generate a set of residual pixels, wherein computing the set of reconstructed pixels is based on the set of prediction pixels and the set of residual pixels, wherein the set of reconstructed pixels are used to generate a second set of prediction pixels, wherein each of the s a prediction pixels is located in a first position before each of the second set of prediction pixels is located in a second position according to a prediction order of the intra prediction mode;
quantize the set of residual pixels to generate a set of quantized residual pixels;
de-quantize the set of quantized residual pixels to generate a second set of residual pixels; and
combine the second set of residual pixels with the set of prediction pixels to generate the set of reconstructed pixels,
wherein the plurality of reconstructed pixels comprises an initial set of reconstructed pixels, wherein the initial set of reconstructed pixels is based on an initial set of prediction pixels, wherein the intra prediction mode is used to generate the initial set of prediction pixels based on a set of external reference pixels located in one or more neighboring blocks, wherein the one or more neighboring blocks have been coded before the current block, wherein d(i,j) denotes the set of residual pixels, wherein i and j are integers between 0 and N−1, wherein N is a size of the current block, wherein d″(i,j) denotes the second set of residual pixels, wherein floor identifies a location of the set of residual pixels relative to the second set of residual pixels, and wherein qp_scale denotes a quantization step determined by a quantization parameter (QP), and wherein d″(i,j)=floor(d(i,j)/qp_scale)*qp_scale.

2. The apparatus of claim 1 , wherein the plurality of reconstructed pixels are equal to the plurality of original pixels.

3. The apparatus of claim 1 , wherein the processor is further configured to:
perform a transform on the set of residual pixels to generate a plurality of transform coefficients;
perform an inverse transform on the plurality of transform coefficients to generate a second set of residual pixels; and
combine the second set of residual pixels with the set of prediction pixels to generate the set of reconstructed pixels.

4. The apparatus of claim 3 , wherein the transform and inverse transform are invertible.

5. The apparatus of claim 4 , wherein the transform is performed on a matrix comprising the set of residual pixels, wherein other positions of the matrix different from the set of residual pixels are assigned with arbitrary values, and wherein the arbitrary values are zero pixel values.

6. The apparatus of claim 1 , wherein the current block is a prediction unit (PU).

7. The apparatus of claim 1 , wherein the intra prediction mode is a vertical mode.

8. The apparatus of claim 1 , wherein the processor is further configured to encode the set of residual pixels to generate encoded data.

9. A method comprising:
receiving a video frame comprising a current block, wherein the current block comprises a plurality of original pixels;
computing a plurality of reconstructed pixels based on the original pixels;
using an intra prediction mode to generate sets of prediction pixels in sequence one of vertically and horizontally, but not both, for the current block, wherein the sets of prediction pixels are each smaller than the current block and based on a set of reconstructed pixels in the plurality of reconstructed pixels, and wherein a set of internal reference pixels from within the current block immediately precedes each set of prediction pixels generated in sequence after an initial set of prediction pixels such that a prediction distance remains one pixel regardless of a size of the current block;
selecting the intra prediction mode from a plurality of available intra prediction modes;
computing a difference between each of the set of prediction pixels and each of a corresponding set of original pixels in the plurality of original pixels to generate a set of residual pixels, wherein computing the set of reconstructed pixels is based on the set of prediction pixels and the set of residual pixels, wherein the set of reconstructed pixels are used to generate a second set of prediction pixels, wherein each of the set of reconstructed pixels is located in a first position before each of the second set of prediction pixels is located in a second position according to a prediction order of the intra prediction mode;
quantizing the set of residual pixels to generate a set of quantized residual pixels;
de-quantizing the set of quantized residual pixels to generate a second set of residual pixels; and
combining the second set of residual pixels with the set of prediction pixels to generate the set of reconstructed pixels,
wherein the plurality of reconstructed pixels comprises an initial set of reconstructed pixels, wherein the initial set of reconstructed pixels is based on an initial set of prediction pixels, wherein the intra prediction mode is used to generate the initial set of prediction pixels based on a set of external reference pixels located in one or more neighboring blocks, wherein the one or more neighboring blocks have been coded before the current block, wherein d(i,j) denotes the set of residual pixels, wherein i and j are integers between 0 and N−1, wherein N is a size of the current block, wherein d″(i,j) denotes the second set of residual pixels, wherein floor identifies a location of the set of residual pixels relative to the second set of residual pixels, wherein qp_scale denotes a quantization step determined by a quantization parameter (QP), and wherein d″(i,j)=floor(d(i,j)/qp_scale)*qp_scale.

10. The method of claim 9 , wherein the plurality of reconstructed pixels are equal to the plurality of original pixels.

11. The method of claim 9 , further comprising:
performing a transform on the set of residual pixels to generate a plurality of transform coefficients;
performing an inverse transform on the plurality of transform coefficients to generate a second set of residual pixels; and
combining the second set of residual pixels with the set of prediction pixels to generate the set of reconstructed pixels.

12. The method of claim 11 , wherein the transform and inverse transform are invertible.

13. The method of claim 12 , wherein the transform is performed on a matrix comprising the set of residual pixels, wherein other positions of the matrix different from the set of residual pixels are assigned with arbitrary values, and wh

Assignees

• Gao Wen
• Song Jin
• Yang Mingyuan
• Yu Haoping
• Futurewei Technologies Inc

Inventors

• Gao Wen
• Song Jin
• Yang Mingyuan
• Yu Haoping

Classifications

• H04N19/12

  Selection from among a plurality of transforms or standards, e.g. selection between discrete cosine transform [DCT] and sub-band transform or selection between H.263 and H.264 · CPC title

• H04N19/90 Primary

  using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals · CPC title

• H04N19/154

  Measured or subjectively estimated visual quality after decoding, e.g. measurement of distortion (use of rate-distortion criteria H04N19/147) · CPC title

• H04N19/103

  Selection of coding mode or of prediction mode · CPC title

• H04N19/11

  among a plurality of spatial predictive coding modes · CPC title